refactor: tidy up perf pass

[RostiMelk]

• Hoist effectiveDensityFactor above the cache branch in useLogoSoup (was computed in both paths)
• Drop INV_255 constant, inline the division in measure.ts
• Rename threshSq → contrastDistanceSq for clarity on the * 3 derivation
• Extract static inline style properties to module-level constants in LogoSoup.tsx

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
react-logo-soup	Ready	Preview, Comment	Feb 7, 2026 10:03am

[github-actions]

Benchmark Comparison: main vs refactor/cleanup-perf-pass

Threshold: 5%+ change, >100us absolute delta, and statistically significant (p<0.05).

Benchmark	main	refactor/cleanup-perf-pass	Change	p-value	Verdict
content detection (1 logo)	36.68 us	35.30 us	-3.8%	0.501	unchanged
render pass (20 logos)	869 ns	957 ns	+10.1%	0.001 **	unchanged
mount 20 logos (no detection)	3.20 us	3.37 us	+5.5%	0.536	unchanged
mount 20 logos (defaults)	828.07 us	818.07 us	-1.2%	0.297	unchanged

No regressions detected.

Feature cost breakdown

How expensive are individual features? Measured on this run's HEAD commit.

Feature	On	Off	Cost	Sig
densityAware: true vs false	32.07 us	32.54 us	~same	0.596
alignBy: visual-center-y vs bounds	1.14 us	57 ns	20× slower	<0.001 ***
cropToContent: true vs false	2.15 ms	43 ns	2.15 ms	<0.001 ***
layout update: full mount vs cached	824.03 us	1.75 us	470× slower	<0.001 ***

Full benchmark output in the CI job logs.