Fix TypeError when rowalign is passed as tuple (#440)

[thejedi433]

Problem

Calling tabulate() with rowalign (or any alignment parameter) as a tuple raises TypeError:

from tabulate import tabulate
tabulate([['a', 'b']], rowalign=('center',), tablefmt='grid')
# TypeError: can only concatenate tuple (not "list") to tuple

This was reported in #440.

Root cause

_expand_iterable (line 2482 in tabulate/__init__.py) uses original + [default] * N. When original is a tuple, tuple + list raises TypeError.

def _expand_iterable(original, num_desired, default):
    if isinstance(original, Iterable) and not isinstance(original, str):
        return original + [default] * (num_desired - len(original))  # tuple + list -> boom

Fix

Two changes in _expand_iterable:

1. Convert original to a list before concatenation — handles tuple, set, generator, and any other iterable uniformly.
2. Add bytes to the exclusion list alongside str for consistency (a bytes value shouldn't be expanded element-wise either).

def _expand_iterable(original, num_desired, default):
    if isinstance(original, Iterable) and not isinstance(original, (str, bytes)):
        return list(original) + [default] * (num_desired - len(original))

This affects rowalign, colalign, numalign, stralign, numparse, maxcolwidths, and maxheadercolwidths — all of which flow through _expand_iterable.

Tests

Added test/test_rowalign_tuple.py with 8 tests covering:

• Tuple rowalign no longer raises TypeError
• Tuple and list inputs produce identical output
• Partial tuples (fewer elements than rows) still pad correctly
• Regression tests confirming list, None, and all other alignment parameters (colalign, numalign, stralign) work with tuples

All 314 existing tests still pass (60 skipped due to optional dependencies).

Scope

Minimal — two lines changed in _expand_iterable, no other behavior affected. No documentation or API changes.