@W-23545283 native lib benchmarks

[andres-rad]

What

Adds a language-agnostic benchmark harness for the DataWeave native-lib Node wrapper under a new top-level benchmarks/ directory. It measures four metrics — cold-start, first-run (compile+exec), warm steady-state, and streaming throughput — emits results in a shared JSON schema, and renders a comparison report. Architected so future Python and Scala-engine runners plug in unchanged.

W-23545283.

How it works

• benchmarks/corpus/ — the shared contract every runner consumes: manifest.json (each case has an immutable id, a .dwl script, optional inputs, and an explicit metrics[]), 6 scripts, committed inputs, and a deterministic large-input generator.
• benchmarks/schema/result.schema.json — the JSON schema all runners' output conforms to (env carries weaveVersion/commit/dwlibBuildId for attribution).
• benchmarks/lib/ — dependency-free helpers: stats (percentiles + MB/s), manifest (load/validate + fail-fast orphan-id check), env (env stamp).
• benchmarks/runners/node/ — the reference runner: in-process warm/streaming (warm-bench), a fresh-process spawn harness for cold-start/first-run (coldstart), and emit (orchestrate → validate ids → write result JSON).
• benchmarks/report/report.mjs — joins result files by case id, renders a case×metric table with a Δ-vs-baseline column and a weave-version skew banner; dedupes to the latest result per runner. --emit is a reserved exporter seam.
• native-lib:benchmark Gradle task — opt-in via -Pbenchmark=true, off by default (never in the normal build/test path).

Design decisions

• Timing is a self-contained process.hrtime.bigint() sampling harness (not vitest bench) so a future JVM-engine harness can mirror the exact methodology for a fair cross-surface comparison.
• first-run is measured in the spawn harness (fresh process = genuine cold compilation).
• Results are local-only/gitignored; durable history (Grafana/exporter) is a deferred follow-up, with env.commit/dwlibBuildId + immutable id + the --emit seam in place so it's a bolt-on.

Testing

• 19 harness tests pass (node --test across benchmarks/**/*.test.mjs), including integration tests that drive the real native library.
• Verified the full Gradle path end-to-end on GraalVM (./gradlew native-lib:benchmark -Pbenchmark=true, exit 0) — builds dwlib, runs all four metrics, renders the report. Numbers are reproducible run-over-run (~1–3% variance).

Out of scope (follow-ups)

• Python runner (runners/python/).
• The Scala-engine baseline harness — must live in the data-weave repo (depends on DataWeaveScriptingEngine); reads this corpus + emits this schema. Deserves its own spec (JVM cold-start/JIT-warmup methodology).
• Results datastore / dashboard.

Design spec and implementation plan are included under docs/superpowers/.