Fused modulo-equality lane; admit const bindings to comparison-lane reads

[lahma]

Rebased onto latest main after #2603 merged.

What

Two commits, each independently measured:

1. Fused modulo-equality lane: identifier % numericConstant ==/===/!=/!== numericConstant —
   the stopwatch.js if-chain shape (z % 2 == 0, executed ~820k times per benchmark op) — is
   evaluated as one fused lane on the equality expressions: the identifier is read through the same
   slot/global arms as the comparison lane and the whole test runs on raw doubles, skipping the
   modulo node's dispatch, operand materialization and result boxing.
2. Admit immutable bindings to lane reads: TryGetNumberSlot requires binding.Mutable so its
   read-modify-write callers can pair it with SetNumberSlot unchecked — but the comparison lanes
   reused it for pure reads, inheriting the write-side constraint. Consequence: no lane ever fired
   on a const operand (stopwatch-modern's entire const z chain declined per evaluation). The
   lanes now read through TryGetNumberSlotForRead, a read-only sibling without the gate.

Semantics

C# double % implements Number::remainder exactly; the only integer-vs-double remainder
divergence (−0 vs +0 results) is erased by the equality consumer (IEEE == treats them equal) — so
the fused result is spec-exact for any proven-Number operand: NaN dividends, % 0 and NaN constants
compare false, v % ±Infinity == v. TDZ stays intact for the const admission: an uninitialized
binding has neither an unboxed number nor a reference value, so it declines and the generic path
raises the ReferenceError. Writers keep the Mutable-gated form (compound assignment on a const still
throws through the generic path).

Results

Commit 1 (fused lane), A/B vs the global-arm base (default job, adjacent runs):

row	before	after
LoopDispatch.ModuloEqualTest	8.05 ms / 2,808 KB	5.56 ms / 3.3 KB	−31% time, −99.9% alloc
stopwatch driver	167.9 ms/iter	152.6 ms/iter	−9.1%
StopwatchBenchmark classic Execute	150.0 ms	136.5 ms	−9.0%
StopwatchBenchmark classic Prepared	158.4 ms	141.8 ms	−10.5%
stopwatch-modern driver	183.9 ms/iter	187.5 ms/iter	+2.0% — the const gap, fixed by commit 2

Commit 2 (const admission), cumulative A/B vs the same base:

row	before	after
StopwatchBenchmark classic Execute	158.6 ms	141.6 ms	−10.7%
StopwatchBenchmark classic Prepared	152.9 ms	137.5 ms	−10.1%
StopwatchBenchmark modern Execute	173.7 ms	153.0 ms	−11.9%
StopwatchBenchmark modern Prepared	169.7 ms	149.8 ms	−11.7%
stopwatch-modern driver	186.3 ms/iter	167.7 ms/iter	−10.0% (was +2.0% before commit 2)
stopwatch classic driver	167.1 ms/iter	153.0 ms/iter	−8.4%

Allocations flat on every row.

Tests

Fused-lane sign/zero behavior across negative dividends, NaN / % 0 / Infinity specials, the
coercing decline, the global-binding shape, const bindings through all three lane forms, and the TDZ
ReferenceError.

Gating

All-TFM build ✅ · Jint.Tests ×2 ✅ · PublicInterface ×2 ✅ · Test262 99,260/0 ✅

🤖 Generated with Claude Code