Finish the repository handoff cleanup

[dv00d00]

Summary

• repair the WHATWG generators after the source layout changes and verify live regeneration against the checked-in tables
• make benchmark tiers unambiguous, expose the UTF-8 corpus benchmark explicitly, and preserve the canonical 47-page published comparison set
• remove unused retained/compact DOM helpers and reuse the shared HTML class-token implementation
• split compact node text operations, tokenizer raw/script modes, query plan pipelines/rewriting, and rewrite collector implementation into cohesive files

The structural commits only move existing method bodies. The branch changes 21 files with a net increase of 23 lines.

Validation

• repository-local CSharpier: 242 / 242 files
• full Release solution build: 0 warnings, 0 errors
• net10.0 tests: 179,562 / 179,562 passed
• UTF-8 tokenizer smoke: 57 documents / 38,831,169 bytes, 0 failures
• UTF-8 DOM projection and QQ / QQ-x4 scraper checks passed
• live WHATWG regeneration: 228 encoding labels and 2,231 entities, byte-for-byte equal to checked-in output
• PowerShell syntax, exact BenchmarkDotNet filter counts, and canonical corpus contract checked
• all moved tokenizer/query ranges compared byte-for-byte with main