feat: support reading existing chapters from MKV files

[xMohnad]

Add mkvextract-based parsing so MKVFile can expose chapters that
already exist in a source file, instead of only supporting chapter
creation.

Closes #125

[GitBib]

blocker: this changes mux output for every file that already has chapters, it's not just a read API.

chapters_obj gets populated in __init__ (MKVFile.py:259), and command() at MKVFile.py:393 emits --chapters <tmp.xml> whenever chapters_obj is set. So chapters that mkvmerge used to copy natively now round-trip through Chapters, and whatever the struct doesn't model is dropped.

repro:

cat > chap.xml <<'X'
<?xml version="1.0"?>
<Chapters><EditionEntry><EditionUID>1111</EditionUID><ChapterAtom>
  <ChapterUID>2001</ChapterUID>
  <ChapterTimeStart>00:00:00.000000000</ChapterTimeStart>
  <ChapterPhysicalEquiv>10</ChapterPhysicalEquiv>
  <ChapterDisplay><ChapterString>Intro</ChapterString>
    <ChapterLanguage>chi</ChapterLanguage>
    <ChapLanguageIETF>zh-Hans</ChapLanguageIETF></ChapterDisplay>
</ChapterAtom></EditionEntry></Chapters>
X
mkvmerge -o src.mkv --chapters chap.xml tests/file_2.mkv
python -c "from pymkv import MKVFile; MKVFile('src.mkv').mux('out.mkv', silent=True)"
mkvextract chapters out.mkv

field	source	master	this PR
ChapterPhysicalEquiv	10	10	dropped
ChapLanguageIETF	zh-Hans	zh-Hans	zh

Chapters covers EditionUID/hidden/default/ordered plus atom start/end/uid/hidden/enabled/display. Matroska chapter XML has more — ChapterSegmentUID, ChapterPhysicalEquiv, ChapterTrack, ChapterProcess, EditionDisplay. All of it disappears on remux now, silently.

reading chapters shouldn't touch the write path. suggestion: record whether chapters came from the source file or were set by the caller, and only emit --chapters in the second case. otherwise leave pass-through to mkvmerge.

two more:

1. the integration test never exercises the feature. both fixtures have zero chapters:

$ mkvmerge -J tests/file.mkv | jq .chapters
[]

so test_init_populates_chapters_obj_from_existing_file always takes the else branch and asserts chapters_obj is None. it needs a file with chapters — build one in the test with mkvmerge --chapters.

2. the sp.run mocks in test_mkv_chapters_tags.py are what hid both of the above. hand-written XML has no BOM, no ChapLanguageIETF, and no ChapterCountry that mkvextract derives on its own. mkvtoolnix is installed in CI, so run against real mkvextract output.

the parser itself is fine — parse_chapters_xml ate real mkvextract output, BOM included, without complaint.

the macos failure on this PR is unrelated — the fixture host was serving a corrupt file. fixed in #127, rebase on master and CI should go green.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.30%. Comparing base (acf00ae) to head (f12c9dc).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #126      +/-   ##
==========================================
+ Coverage   99.12%   99.30%   +0.17%     
==========================================
  Files          21       21              
  Lines        1487     1575      +88     
==========================================
+ Hits         1474     1564      +90     
+ Misses         13       11       -2     

☔ View full report in Codecov by Harness.
📢 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.