feat: add isolated Wurst benchmark mode

[Donach]

Adds a first-class isolated JVM benchmark worker mode to WurstScript.

• validates package-level, parameterless @benchmark functions returning int;
• allows one function to be both @test and @benchmark, while keeping @compiletime incompatible;
• discovers benchmarks separately and runs every measured fork in a fresh compiler JVM;
• performs initialization, warmup, calibration, and measured samples in a stable order;
• emits the private wurst-benchmark-worker-v2 contract with identity, checksum, batch size, and raw nanosecond samples only;
• rejects test-only -testFilter and -testTimeout options in benchmark mode;
• closes compile-time and native resources on both success and failure.

Verification from nix develop /home/daniel/Repositories/tever/tever_main on the final tree:

• rtk gradlew test — full compiler suite passed;
• rtk gradlew shadowJar — passed;
• companion Grill integration — stdlib typecheck passed, full quiet stdlib test suite passed, and targeted PolygonTests passed 22/22 with 0 compiler errors and 0 warnings;
• real Polygon comparison — 5 forks, 5 warmups, 20 measured iterations per fork, matching checksum 7907.

The final benchmark measured a 396,165,032 ns accelerated median versus 3,077,294,370 ns linear median: 7.77× median speedup. These are relative JVM-hosted Wurst IL interpreter measurements, not Warcraft III runtime numbers.

Companion CLI/documentation PR: wurstscript/WurstSetup#75. The built-in annotation declaration is proposed separately in wurstscript/WurstStdlib2.

[Donach]

Final controller-generated benchmark evidence from the exact content trees later squashed into the PR commits (tree hashes were checked unchanged after squash).

• Command: grill benchmark Polygon --forks 5 --warmup 5 --iterations 20 --format json
• Accelerated median: 396,165,032 ns/op
• Linear median: 3,077,294,370 ns/op
• Median speedup: 7.7677x
• Both checksums: 7907
• Samples: 100 per implementation across 5 isolated JVM forks
• Compiler JAR: sha256:830c5a70409b76af8fa2a90889a4604712d5fba443e5036d1eea9af94aa9dd8b
• Raw artifact SHA-256: f8cb89840a0d60d3ed7447a98afbd1f0d9ffa24810f86c494ae502eee0ec1d1f

Raw JSON artifact

{
  "schema": "wurst-benchmark-v1",
  "disclaimer": "Benchmark results measure the JVM-hosted Wurst IL interpreter. They vary with the machine, JVM, compiler version, host load, and benchmark setup. Use them for controlled side-by-side comparisons under the same conditions, not as absolute Warcraft III, Jass, Lua, or in-game performance numbers.",
  "environment": {
    "os": "Linux",
    "jvm": "25.0.2",
    "compiler": "sha256:830c5a70409b76af8fa2a90889a4604712d5fba443e5036d1eea9af94aa9dd8b",
    "grill": "1.4.0.0-nightly-master-4-g46b2d02",
    "cpuCount": 16
  },
  "filter": "Polygon",
  "forks": 5,
  "warmup": 5,
  "iterations": 20,
  "benchmarks": [
    {
      "qualifiedName": "PolygonTests.benchmarkAcceleratedClassification",
      "checksum": 7907,
      "forks": [
        {
          "fork": 1,
          "batchSize": 1,
          "samplesNanos": [
            368695448,
            388876391,
            359626453,
            395386916,
            370663259,
            411735192,
            360463508,
            386470017,
            364758746,
            394589097,
            364757655,
            383784729,
            369970288,
            381094280,
            362168867,
            402393185,
            364425923,
            388975482,
            367942356,
            405378551
          ]
        },
        {
          "fork": 2,
          "batchSize": 1,
          "samplesNanos": [
            390771854,
            458847486,
            463081227,
            510428099,
            372039055,
            414835844,
            374836090,
            409322444,
            381503712,
            424050419,
            377278481,
            416840215,
            386946866,
            426031436,
            383267820,
            410637414,
            375440797,
            412517585,
            388279435,
            387815615
          ]
        },
        {
          "fork": 3,
          "batchSize": 1,
          "samplesNanos": [
            384784109,
            428797330,
            402498415,
            465103517,
            386108011,
            423878216,
            385825022,
            433709680,
            396165032,
            420415589,
            409150086,
            427668683,
            392532353,
            429282035,
            406411227,
            415995627,
            394109590,
            418071658,
            409705249,
            405370088
          ]
        },
        {
          "fork": 4,
          "batchSize": 1,
          "samplesNanos": [
            414072451,
            396246406,
            418434750,
            390750933,
            418060587,
            398613829,
            412787230,
            394409881,
            414325553,
            402646238,
            431413905,
            401111636,
            398001895,
            414546813,
            401323076,
            422346471,
            400226547,
            430628003,
            398766366,
            428957415
          ]
        },
        {
          "fork": 5,
          "batchSize": 1,
          "samplesNanos": [
            369997541,
            386462812,
            369566293,
            408023738,
            372444783,
            406191666,
            369961088,
            407032128,
            379757878,
            403177396,
            361511843,
            394286881,
            365488847,
            428558442,
            390979809,
            380326681,
            385429100,
            370491291,
            387502066,
            371115225
          ]
        }
      ],
      "statistics": {
        "mean": 398884852.47,
        "standardDeviation": 25354460.413373582,
        "min": 359626453,
        "max": 510428099,
        "median": 396165032,
        "p90": 428558442,
        "p95": 431413905
      }
    },
    {
      "qualifiedName": "PolygonTests.benchmarkLinearClassification",
      "checksum": 7907,
      "forks": [
        {
          "fork": 1,
          "batchSize": 1,
          "samplesNanos": [
            3132902675,
            3067167963,
            3166785904,
            3182266664,
            3114694566,
            3203361512,
            3162264104,
            3138459019,
            3133064619,
            3164758554,
            3109830442,
            3139697189,
            3173653490,
            3078734573,
            3117062534,
            3239393029,
            3117036597,
            3107556801,
            3127308615,
            3103609486
          ]
        },
        {
          "fork": 2,
          "batchSize": 1,
          "samplesNanos": [
            3071094678,
            3064336179,
            3098448663,
            3073401196,
            3172979183,
            3038884731,
            3070162844,
            3037044835,
            3039965731,
            3370020905,
            3082659881,
            3077294370,
            3085079658,
            3094553551,
            3025968134,
            3073774991,
            3059496017,
            3067219642,
            3085895939,
            3087553825
          ]
        },
        {
          "fork": 3,
          "batchSize": 1,
          "samplesNanos": [
            3153603425,
            2983840982,
            2964823825,
            2959004012,
            2974845813,
            2936944315,
            3008420737,
            2981190371,
            2968133714,
            2959879918,
            3108065743,
            3026574451,
            3049891475,
            3010793584,
            2921572924,
            2928056935,
            2949294249,
            2949800043,
            2952196727,
            2955657770
          ]
        },
        {
          "fork": 4,
          "batchSize": 1,
          "samplesNanos": [
            3016008099,
            3058396342,
            3038951974,
            2988359249,
            3018712963,
            3108180584,
            2978482974,
            3024881381,
            3009479143,
            3044290413,
            3118117305,
            3116563668,
            3039004591,
            3036238531,
            3036610068,
            3121108244,
            3075213605,
            3045029600,
            3020826111,
            3027068145
          ]
        },
        {
          "fork": 5,
          "batchSize": 1,
          "samplesNanos": [
            3137694776,
            3095111533,
            3152670934,
            3136527288,
            3117496025,
            3163545958,
            3244422324,
            3081602943,
            3095713912,
            3133989786,
            3091861231,
            3101855693,
            3067199969,
            3104525899,
            3073945232,
            3066253393,
            3199333699,
            3085322659,
            3085018665,
            3089108087
          ]
        }
      ],
      "statistics": {
        "mean": 3075427572.98,
        "standardDeviation": 75177392.59450755,
        "min": 2921572924,
        "max": 3370020905,
        "median": 3077294370,
        "p90": 3163545958,
        "p95": 3182266664
      }
    }
  ]
}

[Frotty]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d35b2a8fcc

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Reject extension functions annotated as benchmarks

When a user writes an annotated extension function such as @benchmark function int.foo() returns int, the validator does not reject it because extension functions allow annotations and checkBenchmark is only called for FuncDef/NativeFunc; however this FuncDef-only flagging means the function is never given IS_BENCHMARK, so -benchmarkList silently omits it and exact execution reports it as missing. Please either reject ExtensionFuncDef explicitly as not parameterless/package benchmark material, or support invoking it with its receiver.

Useful? React with 👍 / 👎.