test: deflake test_actor_create_update_delete against retried create conflict

[vdusek]

The integration test test_actor_create_update_delete intermittently failed with ConflictError: Some other Actor already has this name, e.g. in this CI run (on an unrelated Renovate PR).

Root cause: the HTTP client retries every request on transient 5xx and network errors (max_retries=8). A POST /v2/actors create can commit server-side on one attempt yet still be retried; the retry then hits a 409 on the unique name it just created. The failing traceback shows attempt=2, and the 61^8 random resource name rules out cross-worker or leftover collisions.

Create calls in the integration tests now go through conflict-recovering helpers in tests/_utils.py. On the API error that signals a lost name or version number, the helper fetches what the first attempt committed and the test continues. Anything else propagates, and so does a matching error whose resource cannot be found afterwards - a genuine create failure still fails the test rather than being masked.

This covers every create of a resource with a uniqueness constraint: Actors (actor-name-not-unique), tasks (actor-task-name-not-unique), schedules (schedule-name-not-unique), Actor versions (version-already-exists) and env vars (env-var-already-exists). Webhooks have no name uniqueness, and the dataset/KVS/queue creates use get_or_create, so neither needs a helper.

The recovery branches only run when a retried create already committed, which never happens in a normal integration run. tests/unit/test_create_helpers.py therefore drives each helper against a mocked API, so a wrong error type or lookup route is caught there instead of resurfacing as the flake the helpers exist to prevent.

This is a test-side fix. The underlying client behaviour - retrying non-idempotent requests - is tracked in #990; besides these spurious conflicts it can also silently duplicate run starts and webhook creations for callers.

✍️ Drafted by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.71%. Comparing base (297a53d) to head (5ef747a).
⚠️ Report is 4 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #971      +/-   ##
==========================================
+ Coverage   94.64%   94.71%   +0.06%     
==========================================
  Files          58       58              
  Lines        5248     5259      +11     
==========================================
+ Hits         4967     4981      +14     
+ Misses        281      278       -3     

Flag	Coverage Δ
integration	92.14% <ø> (-0.41%)	⬇️
unit	84.92% <ø> (+1.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[vdusek]

converting to draft as I'm not sure whether these test utils are worth it, let me @Pijukatel know WDYT