refactor: extract _rewrite_html() to avoid double mistune in rewrite_comment()

[its-me-maady]

rewrite_comment() previously called self.rewrite() which internally called self.markdown() again, causing comment text to go through mistune twice.

Extract the BeautifulSoup half of rewrite() into _rewrite_html(). rewrite() now calls _rewrite_html() after its markdown step. rewrite_comment() calls _rewrite_html() directly, skipping the second markdown pass.

Split the single try/except in rewrite() into two separate ones — one for the markdown step, one for BS4 init in _rewrite_html() — preserving the same error behavior (return content on failure).

Fixes #398

[codecov]

Codecov Report

❌ Patch coverage is 42.85714% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 9.84%. Comparing base (49b1247) to head (23d8953).

Files with missing lines	Patch %	Lines
src/sotoki/utils/html.py	42.85%	4 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##            main    #399      +/-   ##
========================================
- Coverage   9.91%   9.84%   -0.08%     
========================================
  Files         26      26              
  Lines       2441    2439       -2     
  Branches     316     316              
========================================
- Hits         242     240       -2     
- Misses      2186    2188       +2     
+ Partials      13      11       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[benoit74]

I know this was here before this PR, but it is now quite obvious that we should strive to make all these soup / markdown errors fatal, raising an Exception which will stop the scraper. I see little to no reason for these errors to happen, and should they happen it would make more sense to identify and fix the bug rather than creating broken ZIMs without noticing it.

I've checked few production log and I've never seen such errors, so it probably never happens on current dumps. Or it is a rare edge case which is probably worth fixing or detecting better to make more informed decisions in this edge case.

[its-me-maady]

Good point — I've removed all the try/except blocks around the markdown and soup steps and let them raise instead. If these ever fail, it's better to crash loud than silently produce a broken ZIM.

[its-me-maady]

PR is ready for review.

Also wanted to propose an idea: while working on this repo I kept forgetting to run ruff and black --check before committing, and missed adding the CHANGELOG.md entry. Would it make sense to add a git pre-commit hook (or a CI step) that enforces both automatically? Happy to work on that if it sounds like something worth having.