docs: update build performance benchmarks (3.9.2)

[github-actions]

Automated build benchmark update for 3.9.2 from workflow run #689.

[greptile-apps]

Greptile Summary

This is an automated documentation-only PR updating build performance benchmarks for v3.9.2, generated by the CI workflow. It adds new rows to the benchmark history table and expands the README's summary metrics, including a new per-language resolution precision/recall breakdown section.

The updated numbers reveal several notable regressions worth flagging for awareness (not blocking this docs PR):

• Native build speed increased 54% (10.2 → 15.7 ms/file), and WASM is now marginally faster than native at scale
• Overall resolution recall nearly halved (71.1% → 42.9%), which is a substantial regression in edge detection quality
• Resolution precision dropped from 100% to 90.7% for the native engine

Confidence Score: 5/5

Safe to merge — pure automated documentation update with no code changes.

All changes are auto-generated benchmark data produced by CI. Numbers are internally consistent (per-file estimates in README match the raw ms/file figures, percentage deltas in BUILD-BENCHMARKS.md are accurate). The performance regressions surfaced here should be tracked in code, not in this docs PR. No blocking issues.

No files require special attention, though the 54% native build-speed regression and the drop in resolution recall (71.1% → 42.9%) documented here warrant follow-up investigation in the source code.

Vulnerabilities

No security concerns identified.

Important Files Changed

Filename	Overview
README.md	Updated benchmark summary table for v3.9.2 and added a new collapsible per-language resolution precision/recall breakdown; all numbers are consistent with the detailed benchmark file.
generated/benchmarks/BUILD-BENCHMARKS.md	Two new rows added for v3.9.2 (native and WASM engines) with per-file build speed, query time, node/edge density, and DB size metrics; correctly annotated with percentage deltas vs prior version.

Sequence Diagram

sequenceDiagram
    participant CI as CI Workflow (#689)
    participant BM as Benchmark Runner
    participant GH as GitHub Actions Bot
    participant PR as Pull Request #902

    CI->>BM: Trigger benchmark suite (v3.9.2)
    BM->>BM: Run build benchmarks (native + WASM)
    BM->>BM: Run resolution precision/recall tests
    BM->>BM: Compute per-language breakdown
    BM->>GH: Emit benchmark results (JSON + Markdown)
    GH->>PR: Update generated/benchmarks/BUILD-BENCHMARKS.md
    GH->>PR: Update README.md summary table + per-language table
    PR-->>CI: Automated docs PR opened

Loading

_{Reviews (1): Last reviewed commit: "docs: update build performance benchmark..." | Re-trigger Greptile}