@W-23599590 Add CLI runner to DataWeave native benchmark harness

[andres-rad]

Summary

• Adds a 4th benchmark runner (benchmarks/runners/cli/) measuring the shipped dw native CLI binary, completing the cross-surface comparison alongside the Node/Python FFI wrappers and JVM engine baseline
• Introduces a build-gated BenchmarkHarness inside native-cli (enabled only under -Pbenchmark=true; GraalVM tree-shakes it from the production binary when disabled) with a parent/child READY-marker protocol for true cold-start, first-run, and warm metrics
• Registers a benchmarkCli Gradle task auto-discovered by the root benchmarkCompare aggregator — no changes to the schema, report, or aggregator required

Test plan

• ./gradlew native-cli:test — 7 new BenchmarkHarnessTest Scala tests pass
• ./gradlew native-lib:test -PskipNodeTests=false — 2 new JS parity tests (locate.test.mjs, emit.test.mjs) pass without a binary
• ./gradlew native-cli:benchmarkCli -Pbenchmark=true — runs end-to-end, writes benchmarks/results/cli-<ts>.json conforming to schema/result.schema.json
• ./gradlew benchmarkCompare -Pbenchmark=true — unified report includes CLI runner alongside existing three
• ./gradlew build (no -Pbenchmark) — production dw binary contains no reachable benchmark code