PR summary

This PR fixes issue #30706.

Why this change is necessary

Passing a non-string value (e.g., np.array, int, None) to the orientation
parameter of Axes.grouped_bar() previously caused a misleading NumPy error:

ValueError: The truth value of an array with more than one element is ambiguous

This occurred because _api.check_in_list() attempted a membership check on a
non-scalar object, leading to an elementwise boolean comparison.

What problem it solves

This patch ensures that invalid (non-string) orientation values raise a clean,
user-facing ValueError instead of an ambiguous NumPy truth-value error.
This behavior now matches other Matplotlib functions (e.g., barh, stem)
that first validate enum arguments before list membership checks.

Implementation details

Added an early type guard in Axes.grouped_bar() before calling _api.check_in_list():

if not isinstance(orientation, str):
    raise ValueError(f"{orientation!r} is not a valid value for orientation")
_api.check_in_list(["vertical", "horizontal"], orientation=orientation)

Tests

Added a new test test_grouped_bar_orientation_invalid() to
lib/matplotlib/tests/test_axes.py verifying the following cases:

"invalid"
1
None
np.array([1, 2, 3])

Each now raises:

ValueError: '' is not a valid value for orientation

Result

Clean ValueError instead of ambiguous truth-value error
Consistent behavior with other Matplotlib APIs
All tests pass locally

PR checklist

closes #30706
new and changed code is tested
[N/A] Plotting related features are demonstrated in an example
[N/A] New Features and API Changes are noted with a directive and release note
[N/A] Documentation complies with general and docstring guidelines