fix: Return nulls for is_finite, is_infinite, and is_nan when dtype is pl.Null

[mcrumiller]

Closes #21238.

Not sure if this is desired behavior, but it was an easy fix, so if so, here's the PR.

import polars as pl

pl.Series("a", [None, None]).is_nan()  # also is_finite and is_infinite
# shape: (2,)
# Series: 'a' [bool]
# [
#         null
#         null
# ]

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.87%. Comparing base (515c1b8) to head (3d81723).
Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #21253      +/-   ##
==========================================
+ Coverage   79.80%   79.87%   +0.07%     
==========================================
  Files        1596     1596              
  Lines      228468   228471       +3     
  Branches     2607     2607              
==========================================
+ Hits       182318   182489     +171     
+ Misses      45554    45386     -168     
  Partials      596      596              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ritchie46]

If the nulls propagate in floats, it should return null. Otherwise it should return false.

If there is propagation, we must ensure this happens for the other dtypes as well.

[mcrumiller]

@ritchie46 nulls do propagate:

>>> import polars as pl
>>> pl.Series([1.0, None]).is_finite()
shape: (2,)
Series: '' [bool]
[
        true
        null
]
>>> pl.Series([1.0, None]).is_infinite()
shape: (2,)
Series: '' [bool]
[
        false
        null
]
>>> pl.Series([1.0, None]).is_nan()
shape: (2,)
Series: '' [bool]
[
        false
        null
]

So this is the correct behavior.