BUG: fix dirint KeyError with scalar inputs on pandas >= 2.0

[Omesh37]

Closes #2751

Description

pvlib.irradiance.dirint crashed with KeyError: 'cannot use a single bool to index into setitem' when called with scalar inputs on pandas >= 2.0.

In _dirint_bins, scalar inputs like solar_zenith=90 cause comparisons such as (zenith >= 0) & (zenith < 25) to return a plain Python bool. pandas >= 2.0 treats scalar bools as label lookups on setitem, raising KeyError.

Fix: Convert all four scalar-or-array inputs to pd.Series at the top of _dirint_bins, ensuring boolean masks are always a boolean Series.

Checklist

• Closes BUG: irradiance.dirint raises KeyError with scalar inputs on pandas >= 2.0 #2751
• I am familiar with the contributing guidelines
• I attest that all AI-generated material has been vetted for accuracy and is in compliance with the pvlib license
• Tests added
• Adds description and name entries in the appropriate "what's new" file in docs/sphinx/source/whatsnew
• New code is fully documented
• Pull request is nearly complete and ready for detailed review

[Omesh37]

Updated the PR based on your feedback:

1. Added scalar_input = np.isscalar(solar_zenith) at the top of
   dirint before any computation
2. Returns float(dni.iloc[0]) if scalar input, pd.Series otherwise
3. Updated docstring: array-like → numeric for ghi and
   solar_zenith, return type updated to numeric or pd.Series
4. Fix in _dirint_bins: scalar inputs cast to pd.Series using
   times index so boolean masks always produce a boolean Series
5. Test added covering:
   • scalar int → float output
   • scalar float with use_delta_kt_prime=False → positive float
   • Series input → Series output (regression)
6. Whatsnew entry added in v0.15.2.rst

All 9 dirint tests pass locally. Please let me know if anything needs adjusting.

[RDaxini]

I don't think the test failure is related to the changes in this PR

[Omesh37]

I don't think the test failure is related to the changes in this PR

Ya it is related to failing test of test_singlediode_precision function in test_singlediode.py.

Thank You :)

[echedey-ls]

Nice PR @Omesh37 . Two minor comments down below.

[cwhanse]

Let's add a check for data length here.

Suggest

if len(times)<2:
    raise ValueError('The DIRINT model requires at least two times')

We want to add a test to confirm that the error gets raised.

[Omesh37]

Fixed the ValueError is now conditional on use_delta_kt_prime=True since that's the only case requiring 2+ time points (as noted by @adriesse). When use_delta_kt_prime=False, single time point works fine.

Test updated to cover both cases:

• use_delta_kt_prime=True + single point → ValueError
• use_delta_kt_prime=False + single point → returns pd.Series