Guard getFixedNum against a None simpleFirstChild

[serrebidev]

Problem

script_readColumn (NVDA+Ctrl+<n>) crashes on some list controls with:

File ".../columnsReview/__init__.py", line 793, in getFixedNum
    startNum = child.columnNumber-1
AttributeError: 'NoneType' object has no attribute 'columnNumber'

getFixedNum reads curItem.simpleFirstChild.columnNumber to work out the invisible-column offset, but simpleFirstChild can be None even when childCount > 0.

Reproduction

Reproducible on a wxWidgets wxListCtrl (report view) on recent wxWidgets 3.2.x builds (via wxPython): NVDA reports the focused list item has columns (childCount > 0) but exposes no simple first child, so every row of the list fails with the error above. (Originally reported against a wxPython app, NVDA 2026.2beta.)

Fix

When there is no usable first child, fall back to treating the pressed number as the 1-based column index instead of raising. The add-on then degrades gracefully to a normal column read (or the existing "no more visible columns" path) rather than an error dialog. No behavior change for lists that already expose a first child.

[serrebidev]

The failing build addon check on this PR is not caused by these changes — it's pre-existing and fails identically on master.

I checked out master at 42d55a9 (no changes from this PR) and ran the CI step locally: SKIP=no-commit-to-branch pre-commit run --all-files fails there with the same 16 ruff errors and 34 pyright errors. The last five build addon runs have all failed for the same reason. None of the reported files (compat.py, installTasks.py, and the bare-except: sites) are touched by this PR — the only change here is the getFixedNum guard.

I've opened #51, which fixes the check itself; it builds green. If you'd like, I'm happy to rebase this PR on top of that once you've decided what to do with it.

The bug this fixes is unrelated to any of that: simpleFirstChild can be None while childCount > 0, so script_readColumn raises AttributeError on every row of an affected wxListCtrl.

[ABuffEr]

Hi,
sorry for delay, and thanks to your contribute!
I'll merge also the other PR, then I'll release a new version in a couple on next week.

[ABuffEr]

Hi,
sorry, I'll reverted this PR, and merged #51.
I'm not familiar with this situation; unless I did something wrong, you should be able to rebase this PR and then I'll merge the fix of getFixedNum.
Thanks for patience.

[ABuffEr]

Hi,
I manually implemented the changes from this PR.
All aligned now.