[SizzleUnrlsd]

Summary

Documentation only: the last box of the roadmap, incremental parsing with Tree-sitter, is retired by decision rather than delivered.

• docs/architecture.md §3.1 gets a dated decision note: the frontend is the standard-library ast module behind the build_hir seam the section already requires, the architecture tests enforce that no layer above the frontend touches a parser object, and every later phase was delivered against that frontend. Tree-sitter would add a native dependency, a second adapter to keep identical to the first, and incremental parsing whose benefit the engine already gets from the per-module persistent cache (§11), an edit re-analysing that module and its importers only, with the whole suite running in about two seconds.
• The pipeline diagrams, the "only once" list, the plugin example and the Phase 1 / Phase 10 lists of §38 name the ast frontend and point at the note; the seam stays, so a Tree-sitter adapter producing the same PyHIR remains possible if error recovery on broken files or sub-second editor feedback ever becomes a requirement.

No code changes.

Closes #41