Failing tests `test_wrap_multiword_non_wide*` in current Pythons

[musicinmybrain]

In Python 3.13.11, 3.14.2, and 3.15.0a2 (but not 3.13.9 or 3.14.0), tests test_wrap_multiword_non_wide and test_wrap_multiword_non_wide_with_hypens fail. This is true for the latest PyPI release 0.9.0 and for the current master, 74885be.

$ python3.14 --version
Python 3.14.2
$ git clone https://github.com/musicinmybrain/python-tabulate.git
$ cd python-tabulate
$ tox -e py314
[…]
========================================================================================= FAILURES =========================================================================================
_______________________________________________________________________________ test_wrap_multiword_non_wide _______________________________________________________________________________

    def test_wrap_multiword_non_wide():
        """TextWrapper: non-wide character regression tests"""
        data = "this is a test string for regression splitting"
        for width in range(1, len(data)):
            orig = OTW(width=width)
            cust = CTW(width=width)

>           assert orig.wrap(data) == cust.wrap(
                data
            ), "Failure on non-wide char multiword regression check for width " + str(width)
E           AssertionError: Failure on non-wide char multiword regression check for width 2
E           assert ['th', 'is', ...e', 'st', ...] == ['th', 'is', ...e', 'st', ...]
E             
E             At index 3 diff: 'a' != 'a '
E             
E             Full diff:
E               [
E                   'th',
E                   'is',...
E             
E             ...Full output truncated (24 lines hidden), use '-vv' to show

test/test_textwrapper.py:18: AssertionError
_________________________________________________________________________ test_wrap_multiword_non_wide_with_hypens _________________________________________________________________________

    def test_wrap_multiword_non_wide_with_hypens():
        """TextWrapper: non-wide character regression tests that contain hyphens"""
        data = "how should-we-split-this non-sense string that-has-lots-of-hypens"
        for width in range(1, len(data)):
            orig = OTW(width=width)
            cust = CTW(width=width)

>           assert orig.wrap(data) == cust.wrap(
                data
            ), "Failure on non-wide char hyphen regression check for width " + str(width)
E           AssertionError: Failure on non-wide char hyphen regression check for width 2
E           assert ['ho', 'w', '...d', '-w', ...] == ['ho', 'w ', ...d', '-w', ...]
E             
E             At index 1 diff: 'w' != 'w '
E             
E             Full diff:
E               [
E                   'ho',
E             -     'w ',...
E             
E             ...Full output truncated (35 lines hidden), use '-vv' to show

test/test_textwrapper.py:30: AssertionError
================================================================================= short test summary info ==================================================================================
FAILED test/test_textwrapper.py::test_wrap_multiword_non_wide - AssertionError: Failure on non-wide char multiword regression check for width 2
FAILED test/test_textwrapper.py::test_wrap_multiword_non_wide_with_hypens - AssertionError: Failure on non-wide char hyphen regression check for width 2
======================================================================== 2 failed, 289 passed, 41 skipped in 0.63s =========================================================================
py314: exit 1 (0.78 seconds) /home/ben/src/forks/python-tabulate> pytest -v --doctest-modules --ignore benchmark pid=1934098
  py314: FAIL code 1 (2.68=setup[1.90]+cmd[0.78] seconds)
  evaluation failed :( (2.72 seconds)

Since this issue is related to non-wide characters, #387 is not a fix for it.