Keep normalized whitespace in sync after filtering

[JunaidAhamed-7777]

Fixes #547.

StripWhitespaceFilter changes whitespace token values but leaves their cached normalized values unchanged. A tab collapsed to a space consequently fails token.match(token.ttype, ' '), and regex matching also sees the original text.

Update normalized when this filter changes the whitespace value. Regression tests cover leading/repeated whitespace, parentheses, and identifier lists, checking both literal and regex matching. All three cases fail before the fix.

• Tests: python -m pytest tests -q --tb=short --basetemp=../tmp-sqlparse on Python 3.14: 509 passed, 2 xfailed, 1 xpassed.
• Ruff: unavailable locally; CI pending.
• Regression tests included.
• No public API or documentation change needed.

git diff --check passes.

Baseline comparison: unchanged base 60cdc649726bf1bc4f1b336050560b336da715ec reports 506 passed, 2 xfailed, 1 xpassed. Both base and final code unexpectedly pass tests/test_regressions.py::test_issue484_comments_and_newlines; the three additional passing cases are the new whitespace regressions.

AI-assisted using GPT-6; human review follows draft submission.