Fixed a corner case where numpy's np.float32 nans are not ignored when using ignore_nan_equality

[noamgot]

This PR fixes the following case:

DeepDiff(np.array([1, 2, 3, np.nan], dtype=np.float32), np.array([1, 2, 4, np.nan], dtype=np.float32), ignore_nan_inequality=True)
# Outputs:
{'values_changed': {'root[2]': {'new_value': 4.0, 'old_value': 3.0},
  'root[3]': {'new_value': nan, 'old_value': nan}}}

As you can see, when there are other differences and we are using np.float32 typed arrays, the np.nan values are not ignored, despite setting ignore_nan_inequality=True.

Notice that in case of the default np.float64 type, it doesn't happen:

DeepDiff(np.array([1, 2, 3, np.nan]), np.array([1, 2, 4, np.nan]), ignore_nan_inequality=True)
# Outputs:
{'values_changed': {'root[2]': {'new_value': 4.0, 'old_value': 3.0}}}

Also, no diffs are observed when there are no other differences:

DeepDiff(np.array([1, 2, 3, np.nan], dtype=np.float32), np.array([1, 2, 3, np.nan], dtype=np.float32), ignore_nan_inequality=True)
# Outputs:
{}

So I think I made it clear that this corner case has not been mitigated properly.
This is my first PR, so please tell me if there's anything I can improve...
Thanks!

[seperman]

LGTM!

[codecov-commenter]

Codecov Report

Merging #376 (4cb8400) into dev (8ab1c8d) will decrease coverage by 0.19%.
The diff coverage is 100.00%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

@@            Coverage Diff             @@
##              dev     #376      +/-   ##
==========================================
- Coverage   99.24%   99.06%   -0.19%     
==========================================
  Files          14       14              
  Lines        3161     3192      +31     
==========================================
+ Hits         3137     3162      +25     
- Misses         24       30       +6     

Impacted Files	Coverage Δ
deepdiff/diff.py	99.63% <100.00%> (+<0.01%)	⬆️
deepdiff/helper.py	98.73% <100.00%> (+<0.01%)	⬆️
deepdiff/commands.py	95.68% <0.00%> (-4.32%)	⬇️
deepdiff/serialization.py	97.36% <0.00%> (-0.41%)	⬇️
deepdiff/model.py	99.75% <0.00%> (+<0.01%)	⬆️
deepdiff/operator.py	96.42% <0.00%> (+1.19%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[seperman]

I will keep you posted when it is published. Thanks!

[noamgot]

Thanks @seperman!