refactor: consolidate codepoint parsing into reusable module

[mit-d]

Summary

• Extract _parse_codepoint() and _parse_range() from main.py into a new check_unicode.parsing module with public
  API
• Add input validation: Unicode range bounds (0..U+10FFFF), empty/invalid input rejection, inverted range detection (lo

hi), whitespace tolerance around range dashes

• Functions raise ValueError instead of argparse.ArgumentTypeError so they're reusable from both CLI and config
  layers
• 25 new unit tests in test_parsing.py covering all formats (U+XXXX, 0xXXXX, bare hex), edge cases, and error paths
• Prerequisite for upcoming configurable replacement table feature

Test plan

• All 265 tests pass (pytest --cov=check_unicode --cov-fail-under=80)
• 98% coverage, parsing.py at 100%
• ruff check and mypy --strict clean
• Smoke test: existing CLI behavior unchanged for --allow-codepoint and --allow-range
• CI matrix (Python 3.11-3.14)