Fix EPUB2 HTML charset metadata

[nfvelten]

Summary

• update existing Content-Type meta charset values regardless of attribute order
• emit XHTML-compatible http-equiv charset metadata instead of HTML5 meta charset for .html content documents
• leave no-head HTML untouched instead of prepending metadata before
• defer fixer lock creation until CLI or EPUBFixer startup so focused imports/tests do not leave stale locks

Fixes #1473.

Tests

• uv run --no-project --with pytest --with tabulate --with requests python -m pytest tests/unit/test_ingest_processor_startup.py tests/unit/test_kindle_epub_fixer.py
• uv run --no-project --with ruff ruff check --select I tests/unit/test_kindle_epub_fixer.py
• uv run --no-project --with ruff ruff check --select SIM115 scripts/kindle_epub_fixer.py
• git diff --check