Replace self._pit_samps NaNs with 0.

[drewoldag]

Problem & Solution Description (including issue #)

We take the approach of simply replacing any NaN values with 0s. And also logging a warning message that it's happening.

Code Quality

• My code follows the code style of this project
• I have written unit tests or justified all instances of #pragma: no cover; in the case of a bugfix, a new test that breaks as a result of the bug has been added
• My code contains relevant comments and necessary documentation for future maintainers; the change is reflected in applicable demos/tutorials (with output cleared!) and added/updated docstrings use the NumPy docstring format
• Any breaking changes, described above, are accompanied by backwards compatibility and deprecation warnings

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (8eaabd3) 100.00% compared to head (5189377) 99.95%.
Report is 5 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##              main     #202      +/-   ##
===========================================
- Coverage   100.00%   99.95%   -0.05%     
===========================================
  Files           34       33       -1     
  Lines         2284     2286       +2     
===========================================
+ Hits          2284     2285       +1     
- Misses           0        1       +1     

Flag	Coverage Δ
unittests	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
src/qp/mixmod_pdf.py	100.00% <100.00%> (ø)
src/qp/metrics/pit.py	98.27% <75.00%> (-1.73%)	⬇️

... and 1 file with indirect coverage changes

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

[drewoldag]

Even though this is closed out, @aimalz if you want to weigh in on the mathematically best path forward, I would like to hear your input. Eric proposed (and I agreed with) what seems like a reasonable solution to the issue, but it's not 100% clear if it would be better to handle the NaNs in a different way.