Fix normalization of np.str_ and np.bytes_ types

[yashwantbezawada]

Reference Issues/PRs

Fixes #2800

What does this implement or fix?

When you try to write a DataFrame with np.str_ values, you get this error:

ArcticDbNotYetImplemented: Failed to normalize column 'col' with dtype 'object'. 
Found first non-null value of type '<class 'numpy.str_'>', but only strings, unicode, 
and Timestamps are supported.

The issue is in _accept_array_string() - it checks type(v) in (str, bytes) which does an exact type match. But np.str_ is its own class that inherits from str, so the check fails even though it is basically a string.

Switched to isinstance(v, (str, bytes)) which handles subclasses properly. Same thing for coerce_string_column_to_fixed_length_array() - changed to_type == str to issubclass(to_type, str).

Any other comments?

Quick sanity check that this makes sense:

>>> isinstance(np.str_("hello"), str)
True
>>> isinstance(np.bytes_(b"hello"), bytes)
True

The C++ side already does the right thing - PyUnicode_Check and PyBytes_Check both check the type hierarchy, so np.str_ values work fine once they get past the Python normalization.

Added some tests to cover this.

Checklist

Checklist for code changes...

• Have you updated the relevant docstrings, documentation and copyright notice?
• Is this contribution tested against all ArcticDB's features?
• Do all exceptions introduced raise appropriate error messages?
• Are API changes highlighted in the PR description?
• Is the PR labelled as enhancement or bug so it appears in autogenerated release notes?

[IvoDD]

Thank you for contributing! Looks good! Just one suggestion for an extra test