Optimize object conversion

[pixelrahulnotfound]

Description

optimise _possibly_convert_objects in xarray/core/variable.py by adding a scan for object arrays.
Here is the logic:
• We scan up the first 10,000(in 99% cases it will usually stop at 1 element and if an array is very large we do it to cap the limit so it doesnt lag in the scan) elements of the flattened object array.
• If a datetime-like type (such as datetime, np.datetime64, pd.Timestamp, pd.NaT, etc.) is found, we flag it as datetime-like and stop the scan.
• If a non-null, non-datetime element is found, we immediately determine it is not datetime-like and stop the scan.
• If no conclusive type is found in the first 10,000 elements (e.g., all elements are null) and the array is larger, we fall back to pandas' highly optimized C-compiled pd.api.types.infer_dtype on the entire array.

Checklist

• [✓] Tests added
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

Tests

• python -m pytest -o addopts="" xarray/tests/test_possibly_convert_objects.py — 11
passed
• ruff check xarray/core/variable.py xarray/tests/test_possibly_convert_objects.py —
passed
• git diff --check — passed

AI Disclosure

[✓] This PR contains AI-generated content.
[✓] I have tested any AI-generated content in my PR.
[✓] I take responsibility for any AI-generated content in my PR.
Tools: Google Antigravity.