fix: handle math run with no text child in OMML->LaTeX conversion

[S1MS4]

Summary

Fixes #2188.

• do_r() in omml.py called elm.findtext("./m:t") and iterated over the result directly. When a math run (<m:r>) has no <m:t> text child (e.g. a run that only carries formatting properties), findtext() returns None, and iterating over None raises TypeError: 'NoneType' object is not iterable.
• Because pre_process_docx() pre-processes math for the whole word/document.xml in one pass wrapped in a blanket try/except, this single malformed run silently drops every equation in the document on fallback, with no error surfaced to the caller.
• Fix: elm.findtext(...) or "" so a missing text child is treated as an empty run instead of crashing.

Verified against a real-world .docx with 62 native equations: before the fix, all 62 were silently dropped from the markdown output; after the fix, all render correctly as LaTeX (e.g. $l\_{1}$, $λ\_{1}\geq λ\_{2}\geq …\geq λ\_{k}$).

This is a different root cause from #1979 (math_root.find("oMath") returning None in _convert_omath_to_latex) and #1982 (unsupported function names) — different function (do_r), different failure mode (missing <m:t> on an otherwise valid run).

Test plan

• Added packages/markitdown/tests/test_docx_omml.py with regression tests:
  • a math run with no <m:t> child no longer crashes and yields an empty string
  • a normal run with text still converts correctly
  • a subscript expression where one run has no text still converts correctly (mirrors the real-world case)
• Confirmed tests fail with the pre-fix code (TypeError) and pass with the fix
• Ran pytest packages/markitdown/tests/test_docx_omml.py -v — all passing

[S1MS4]

@microsoft-github-policy-service agree