HTML table import: keep row order, invent no header cells

[dereuromark]

HtmlToDjot decided a table's header by finding the first row containing any <th>, writing that row as the djot header row - which also moved it to the top and wrote every one of its cells as a header. Two defects fall out of that one rule.

Rows were reordered. This is the serious one: it changes reading order, not markup.

<table><tr><td>a</td></tr><tr><td>b</td></tr><tr><th>H</th></tr></table>

imported on master as:

| H |
|---|
| a |
| b |

A data cell was promoted to a header.

<table><tr><th>R</th><td>1</td></tr></table>

imported as | R | 1 | plus a separator row, so 1 came back as <th> - a column header nobody wrote, which is what a screen reader will announce.

The rule now

A row is the header row only when it is the first row and all of its cells are <th>. Every other row is written in place as an ordinary body row. Rows inside <thead> still count as leading rows.

Why this loss and not the other

djot has no per-cell header spelling. A header row is the row immediately before the |---| separator, and that is the whole of what a djot table can express - unlike Carve, whose fix for the same defect (markup-carve/carve-php#1274) could keep each cell's role. So a <th> outside a leading all-header row cannot survive the import, and the only question is which loss to take:

• Dropping the header marker loses an annotation the author wrote.
• Promoting a data cell invents one they did not.

Inventing is worse than losing, and reordering the reader's content is worse than both. The remaining limitation is stated in the converter docs.

Unchanged

A leading all-<th> row, a <thead>/<tbody> split, and a table with no <th> at all all import exactly as before - each now with a test.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.42%. Comparing base (72442e7) to head (f11736e).

Additional details and impacted files

@@             Coverage Diff              @@
##             master     #275      +/-   ##
============================================
- Coverage     92.42%   92.42%   -0.01%     
  Complexity     3682     3682              
============================================
  Files           109      109              
  Lines         10440    10438       -2     
============================================
- Hits           9649     9647       -2     
  Misses          791      791              

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.