[joeblau]

Closes #36

What

trigger.ts now reads the 11 presets from a local src/presets.ts literal instead of importing defaultPatterns from web-haptics. web-haptics is only imported for types there. HapticTarget still uses WebHaptics unchanged. A new drift test (test/presets.test.ts) deep-equals the local table against web-haptics' defaultPatterns, so a dependency bump that changes presets fails CI.

Implemented with the Codex CLI on the shared box and verified by hand.

Benchmark (npm run bench, same box, back-to-back)

Bundle (esbuild, ES2020, React external)	main 78ec2f2	this PR
triggerHaptic gzip bytes	3,313	1,273 (−2,040 B, −62%)
triggerHaptic min bytes	9,110	3,193
triggerHaptic web-haptics bytes	6,440	0
Full ESM gzip bytes	4,514	4,541 (+27 B, duplicated table)
Full CJS gzip bytes	4,873	4,881 (+8 B)
HapticTarget gzip bytes	3,781	3,782

Byte counts are deterministic. Runtime timings did not change beyond noise.

Verification

• npm run lint, typecheck (ESM+CJS+DTS build), coverage (135 tests, 100% statements/branches/functions/lines), size, pack:check pass locally.
• SSR: the preset table is a plain literal. The imperative path no longer evaluates the web-haptics module, which removes a navigator read at module load.
• Behavior: same patterns, same navigator.vibrate output (existing trigger tests unchanged and passing).

Device-testing notes

These are simulated-browser and bundler numbers (Node 24, jsdom/esbuild on a Linux box), not physical-device measurements. No physical device was used. Physical iPhone/Android/iPadOS/VoiceOver certification remains tracked in #9.