Optimize wildcard quad counts

[jeswr]

Summary

• retain the direct lookup paths for bound counts introduced by Optimize bound quad counts #710
• enumerate wildcard index levels with Object.keys and indexed loops
• keep this as a fallback while the equivalent for-in optimization is investigated upstream in V8

Performance

Measured against current main on Node 25.1.0 with 50,000-quad stores. Each result is the median of 15 paired samples after warm-up, alternating execution order between implementations.

Store / query	Before	After	Improvement
sparse / subject	344.22 ms	278.20 ms	18.5%
sparse / full store	252.92 ms	210.13 ms	23.2%
mixed / subject	65.11 ms	47.80 ms	32.0%

Fully bound and subject-plus-predicate paths return before these loops, so they retain the O(1) behavior from #710 without new key-array allocations.

Verification

• complete Jest suite: 6,847 tests passed
• 100% statement, branch, function, and line coverage
• ESLint passed