feat(python): Allow fill_null to be used with torch tensors

[Niko32]

When trying to use df.fill_null with a value that is a torch tensor I have encountered an unexpected error. I traced it back to some code that would assume a value with an attribute item to be a numpy object.

This is a simple fix for this issue.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.80%. Comparing base (addaf83) to head (28e7fc7).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #19120      +/-   ##
==========================================
- Coverage   79.81%   79.80%   -0.01%     
==========================================
  Files        1531     1531              
  Lines      208596   208595       -1     
  Branches     2913     2914       +1     
==========================================
- Hits       166482   166475       -7     
- Misses      41563    41569       +6     
  Partials      551      551              

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