perf: drop per-element wrapper indirection on offset=0 iterator path

[Farenheith]

What

augmentativeIterateIterable previously wrapped every next() call in a () => current() closure even when no skip offset was set. For the common offset=0 case, return finalNext directly as next(), removing one call frame per element for non-array iterables.

Why it's safe

• Skip path (offset > 0) is byte-identical in semantics to before: stable outer wrapper + internal closure-variable swap. (Reassigning iterator.next after first use would be unsafe — consumers like V8's for...of can cache the function reference, so only an internal variable may change.)
• Array path and async path untouched (augmentativeIterate dispatches arrays to a separate function; async already has minimal indirection by design).
• No API or observable behavior change.

Verification

• 53/53 unit tests pass (also via pre-commit hook)
• Interleaved A/B benchmark, original vs patched in one process, 100k elements:

Scenario	Original	Patched	Δ
gen→array addMap	2.03ms	1.81ms	−10.8%
filter+map chain (gen)	2.08ms	2.01ms	−3.8%
skip(1000)+map (array)	4.37ms	4.44ms	+1.6% (noise — array path untouched)

• Correctness: checksums match original on all scenarios incl. skip paths.

Notes

This is the one remaining safe micro-opt from the Aug 2026 perf analysis pass (destructuring, bind-caching and toArray pre-allocation were measured as noise/harmful and ruled out).