fix: add missing Unicode chars to text normalization map

[hannesrudolph]

Summary

Adds 5 missing Unicode character mappings to NORMALIZATION_MAPS.TYPOGRAPHIC in text-normalization.ts so that apply_diff can match files containing these characters.

Problem

When using apply_diff on a Markdown file containing Unicode typographic characters, the tool reported low similarity scores (73-97%) even when the search content appeared visually identical to the file content. This prevented any edits from being applied.

Repro Steps

1. Create a Markdown file containing a middle dot character: > **Status:** 17 of 36 providers migrated · 19 remaining
2. Attempt to use apply_diff with a search block matching that line
3. The tool reports ~73% similarity and fails to match

The root cause is that the LLM generates ASCII equivalents (- for ·, -> for →, etc.) but normalizeString() didn't know how to map the file's Unicode characters to the same ASCII output.

Changes

src/utils/text-normalization.ts — Added 5 entries to NORMALIZATION_MAPS.TYPOGRAPHIC:

Character	Codepoint	Maps To	Name
·	U+00B7	-	Middle dot
→	U+2192	->	Right arrow
←	U+2190	<-	Left arrow
↔	U+2194	<->	Left-right arrow
(invisible)	U+FE0F	(stripped)	Variation selector

src/utils/__tests__/text-normalization.spec.ts — Added 2 test cases covering all new character mappings.

Testing

cd src && npx vitest run utils/__tests__/text-normalization.spec.ts

All 19 tests pass.

---

Important

Adds missing Unicode character mappings to NORMALIZATION_MAPS.TYPOGRAPHIC in text-normalization.ts and updates tests to improve text normalization accuracy.

• Behavior:
  • Adds 5 Unicode character mappings to NORMALIZATION_MAPS.TYPOGRAPHIC in text-normalization.ts for better text normalization.
  • Characters include middle dot, right arrow, left arrow, left-right arrow, and variation selector.
• Testing:
  • Adds 2 test cases in text-normalization.spec.ts to cover new character mappings.
  • All tests pass successfully.

^{This description was created by for a6c46f7. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Rooviewer   See task

Reviewed the 5 new Unicode character mappings in NORMALIZATION_MAPS.TYPOGRAPHIC and the 2 new test cases. No issues found -- the changes are correct, well-scoped, and adequately tested. All 19 tests pass.

_{Mention @roomote in a comment to request specific changes to this pull request or fix all unresolved issues.}