[BUG] Duplicate author folders from name variants — no deduplication at folder level

[deucebucket]

Problem

When multiple sources report slightly different author names, Library Manager creates separate author folders for each variant instead of normalizing to a single canonical name.

Examples from community tester (Merijeek, #140)

Folder 1	Folder 2	Same person?
Alistair Maclean	Alistair McLean	Yes
Craig Alanson	Craig Allenson	Yes (AI hallucination)
James S. A. Corey	James S.A. Corey	Yes (initials formatting)
Rick Yancey	Richard Yancey	Yes (nickname vs legal name)
N.K. Jemisin	N. K. Jemisin	Yes (initials formatting)

Root cause

The path building system (build_new_path() in path_safety.py) treats author names as exact strings. There's no fuzzy matching or deduplication when deciding which existing author folder to use.

Current tooling that EXISTS but doesn't solve this:

• standardize_initials() — handles J.R.R. → J. R. R. but disabled by default (standardize_author_initials config)
• is_garbage_author_match() — uses word-overlap (Jaccard similarity) for validation, but only at 0.2 threshold and not used during folder creation
• calculate_title_similarity() — Jaccard on word sets, can't catch "Maclean" vs "McLean" (different words)

What's needed

When building a rename path, before creating a new author folder:

1. Check existing author folders in the library for fuzzy matches
2. Use Levenshtein distance or phonetic matching (Metaphone/Soundex) on author names
3. If a close match exists (>85% similarity), use the existing folder name
4. Handle common patterns:
   • Nickname variants: Rick/Richard, Bob/Robert, Bill/William
   • Initial formatting: J.R.R. / J. R. R. / JRR
   • Surname spelling: MacLean/McLean/Maclean
   • Punctuation differences: N.K. vs N. K.

Additional context

The standardize_author_initials config flag should probably be enabled by default — there's no good reason to create James S.A. Corey and James S. A. Corey as separate folders.

Related

• [BUG] More Search Errors #140 - Merijeek's report of duplicate author folders
• [FEATURE] Use path info to complete partial Skaldleita results #127 - Use path info to complete partial SL results
• Smart Pre-Processing Pipeline: Validation, Triage, and Corrections #110 - Smart Pre-Processing Pipeline
• deucebucket/skaldleita#78 - Narrator name fuzzy dedup

[deucebucket]

Fixed in PR #144 (merged to develop). Author folder deduplication now uses 3-tier fuzzy matching (exact, standardized initials, SequenceMatcher >= 85%).