Optimize initial indexing

[plumocracy]

Summary

• scan directories concurrently with a bounded worker pool
• batch directory metadata and classification writes in a single transaction per root
• limit embedded file content to high-signal docs/config/schema files
• ignore SVG/media/archive/font assets by default and in fallback filtering for older configs
• update indexing docs and add regressions for high-signal filtering and SVG exclusion

Verification

• cargo fmt
• cargo test index:: --lib -- --nocapture
• cargo test --all
• cargo clippy --all-targets --all-features -- -D warnings