feat: extract @libscope/parsers — standalone format conversion package

[RobertLD]

Summary

Part of #488. Extract src/core/parsers/ into a standalone @libscope/parsers package. Parsers have zero upward imports to the rest of libscope — this is the lowest-risk extraction in the split.

Problem / Motivation

Format parsers (PDF, DOCX, EPUB, PPTX, CSV, JSON, YAML, HTML → text/markdown) are useful independently of semantic search. Today they're buried inside @libscope/core and carry all of core's dependencies along for the ride. Extracting them allows projects that only need format conversion to take a minimal dependency.

Proposed Solution

Move src/core/parsers/ into packages/parsers/src/ with its own package.json. Each parser's format-specific library (pdf-parse, mammoth, epub2, etc.) stays as an optionalDependency so consumers only install what they need.

Acceptance Criteria

• @libscope/parsers builds independently with npm run build — zero imports from @libscope/core or any other @libscope/* package
• All existing parser tests pass when run from the @libscope/parsers package directory
• Each format-specific library (pdf-parse, mammoth, csv-parse, js-yaml, epub2, node-html-markdown, etc.) is declared as an optionalDependency
• @libscope/core depends on @libscope/parsers and all parser imports in core are updated to the new package path
• Existing libscope CLI behaviour is unchanged

Out of Scope

• Adding new parsers or changing parser output format
• Removing the gzip auto-detection logic in pack file I/O (that lives in core/packs.ts, not parsers)

Technical Notes

• src/core/parsers/ currently has zero upward imports — no imports from indexing, search, DB, or providers. Clean extraction.
• The parsers directory contains: markdown.ts, pdf.ts, docx.ts, xlsx.ts, csv.ts, json.ts, yaml.ts, html.ts, epub.ts, pptx.ts
• All parsers share src/logger.ts and src/errors.ts — these will need to be re-exported from @libscope/core or duplicated into a shared internal package