docs: add public bridge API reference

[leehack]

Summary

• Add docs/api.md as the public JavaScript API reference for LlamaWebGpuBridge.
• Link the API reference from README.md and describe the generated declaration asset.
• Document worker/direct runtime behavior, model loading, completion streaming, tokenization, embeddings, multimodal projector APIs, state persistence, metadata, cancellation/disposal, and browser asset requirements.
• Add static contract checks for API docs coverage and malformed Markdown table rows.
• Fix small public API consistency gaps found while documenting: expose workerGenerationStallTimeoutMs, forward declared worker config keys, and regenerate generated JS/declaration assets.

Closes #11

Test Plan

• python3 -m py_compile scripts/verify_state_persistence_api.py scripts/verify_ci_reliability.py scripts/state_persistence_browser_smoke.py
• python3 scripts/verify_state_persistence_api.py
• python3 scripts/verify_ci_reliability.py
• npm run check:js
• git diff --check

Review Notes

• Pre-push independent docs accuracy review found issues around worker timeout semantics, token callback mode differences, worker config forwarding, Markdown union pipes, pre-load context size behavior, state byte transfer wording, and verifier blind spots.
• Addressed those findings before push by correcting docs, adding the missing public config type, forwarding public config keys into worker config, and hardening the docs verifier.

[Copilot]

Pull request overview

This PR adds a first-class public API reference for the browser-facing LlamaWebGpuBridge assets and introduces CI contract checks to keep the docs and generated declaration assets aligned with the actual exported surface.

Changes:

• Added docs/api.md documenting the public JS API, including worker vs direct-runtime behavior and state persistence semantics.
• Linked the API reference and declaration asset (js/llama_webgpu_bridge.d.ts) from README.md.
• Hardened scripts/verify_ci_reliability.py to enforce API-doc coverage and detect malformed Markdown table rows; also exposed/forwarded a few public config keys and regenerated bundled JS / .d.ts outputs.

Reviewed changes

Copilot reviewed 5 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
scripts/verify_ci_reliability.py	Adds API-doc coverage checks and a Markdown table row validator for docs/api.md.
README.md	Links the new public API reference and documents the published .d.ts asset.
js/src/llama_webgpu_bridge.js	Forwards additional public config keys into worker config.
js/src/llama_webgpu_bridge.d.ts	Exposes workerGenerationStallTimeoutMs on the public config type.
js/llama_webgpu_bridge.js	Regenerated bundled bridge output reflecting forwarded config keys.
js/llama_webgpu_bridge.d.ts	Regenerated published declaration asset.
docs/api.md	New public API reference for LlamaWebGpuBridge.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.