bench(ts): add variadic tuple workload

Context

The accepted TypeScript frontend suite now leads across twelve workloads, but tuple coverage is incidental. It does not isolate variadic tuple construction, readonly tuple inference, conditional head/tail extraction, tuple indexing, generic rest parameters, or spread-call checking.

Proposed fair workload

Add one deterministic, dependency-free TypeScript module containing repeated independent tuple families with:

• readonly literal tuples with distinct element shapes;
• generic variadic tuple concatenation and inferred tuple results;
• conditional head and tail extraction with infer and rest elements;
• fixed and indexed reads from the composed tuple;
• generic rest-parameter calls fed by tuple spreads;
• typed structural consumption and exports retaining every family.

All three compilers must receive the exact same source and tsconfig, accept it successfully, and produce no output under --noEmit before timing. Compiler versions remain JavaScript tsc 6.0.3 and native TypeScript 7.0.2.

Acceptance criteria

• Commit the workload before any optimization and do not change it in response to results.
• Keep the shape deterministic, auditable, dependency-free, and distinct from existing workloads.
• Record source size/hash and an interleaved red/green baseline.
• If Home loses, profile the unchanged workload and make only general compiler improvements.
• Run focused tests plus the complete affected package suites.
• Publish a 30-run result in the detailed performance document and README summary.
• Update #416 with commits and results.

Refs #416

Pre-freeze validity gate is complete.

• The generated workload is deterministic: 5,898 lines, 188,399 bytes, SHA-256 78105a6b28024f5ac1dd32cc81d3aa065a3a99a1972927e1880f0a28eec5324f.
• JavaScript TypeScript 6.0.3 (tsc), native TypeScript 7.0.2 (tsgo), and Home now all accept the identical strict/noLib project silently.
• The initial gate correctly exposed missing generic variadic-spread materialization, trailing conditional tuple inference, and recursive infer-binder preservation. Project-level validation also exposed an unsafe object-member borrow during union reduction and a recursive generic-inference cycle; both were fixed generally rather than hidden by the workload.
• Checker regression coverage now includes concat materialization, conditional head/tail slices, and generic rest capture. The full checker suite passes 4,191/4,191 tests.

Correctness fix: a05b922

No timing result has been accepted yet. The next commit freezes this validated source before measurement.

Frozen 5-run baseline (20260827T072451Z, 1 warmup):

Home is currently 2.08× slower than the fastest competitor on the accepted variadic-tuple workload. This is the optimization target; no timing result is being published as a win yet.

Variadic-tuple benchmark round complete.

• correctness: a05b922dd
• frozen workload: b2476af52
• checker scan/index improvements: beca00892
• relation working-set/hash improvements: 735ab4012
• README and detailed report: 1414a7e69

The generated project remained unchanged: 5,898 lines, 188,399 bytes, SHA-256 78105a6b28024f5ac1dd32cc81d3aa065a3a99a1972927e1880f0a28eec5324f. JavaScript TypeScript 6.0.3, native TypeScript 7.0.2, and Home all validate the identical strict/noLib project successfully and silently.

The final snapshot retains all 30 unfiltered samples and discloses elevated background workstation variance. Compiler order rotated within every round. Home improved 1.36× on the unchanged workload and records lower mean time than both competitors; the accepted suite is now 13/13.

Validation: all 4,191 checker tests pass, the ReleaseFast compiler builds, Pickier reports no errors in the touched docs/code, and every benchmark validity gate is silent.

Refs #416