Wiki links: Add fuzzy matching strategy (middle ground between grep and AI)

[dpshade]

Problem

Currently wiki link identification has two strategies:

• Existing notes only (fast): Simple string matching - instant but only finds exact matches
• All entities via AI (slow): Full semantic analysis - powerful but slow and expensive

There's no middle ground for users who want smarter matching without the AI overhead.

Proposed Solution

Add a third strategy using fuzzy/approximate matching:

Options to explore:

• fzf-style fuzzy matching: Match note titles with typo tolerance and partial matches
• Tantivy: Rust-based full-text search engine, very fast
• MiniSearch: Lightweight JS full-text search
• Fuse.js: Fuzzy search library

Benefits:

• Catch common variations (e.g., "JavaScript" matches "Javascript" or "JS")
• Handle plurals and minor spelling differences
• Still instant/near-instant performance
• No API calls required

Implementation Notes

• Add new WikiLinkStrategy option: "fuzzy"
• Could use similarity threshold setting (e.g., 0.8 = 80% match required)
• Consider indexing note titles on vault load for performance

Related

This would complement the existing strategies without replacing them.