SEO Crawl Audit 0.6.0 architecture and reliability

[MelnixDev]

Summary

• add reusable two-phase scan planning and deterministic breadth-first crawling
• move filesystem persistence behind the core/node subpath and harden durable checkpoints
• split current and regression rule pipelines while preserving rule IDs and fingerprints
• make partial diffs coverage-aware and preserve partial CLI results on signals
• narrow the public core API, split typed CLI commands, and isolate the Action runner
• add strict TypeScript, lint boundaries, coverage gates, package smoke tests, and 9-platform compatibility CI
• document the public API, architecture, cancellation, checkpoints, and Action tag policy

Compatibility

SnapshotV2, SnapshotV1 migration, existing fingerprints, CLI flags, normal exit codes, config precedence, and HTML/JSON outputs remain compatible. The only new CLI exit code is 130 for a correctly interrupted scan.

Verification

• 62 tests pass
• core coverage: 91.26% lines, 80.97% branches, 91.71% functions
• CLI and Action coverage: above 80% lines, 65% branches, and 75% functions
• packed core root/node imports, declarations, and packed CLI pass in a clean temporary install
• production npm audit: 0 vulnerabilities
• CLI bundle: 398,290 bytes; Action bundle: 958,704 bytes
• 10k-page memory and time gate passes

Please merge with a merge commit and do not squash so the logical commit history is preserved.