JP-3250: Exclude entire var if any NaN present

[tapastro]

Resolves JP-3250
Closes #7647

This PR addresses an issue found when a variance array contains NaN values - the sum with the other variance arrays leads to NaN values in the error arrays. The current behavior will propagate any NaNs in a variance array to the final error array.

The approach in this PR is to fully exclude a variance array if at least one NaN is present - this is to prevent the error array from having a variable number of contributions. Alternatively, one could let the error array exclude the NaN values from summation for individual variance entries, leading to the error having 0-3 variance contributions depending on the number of NaNs present.

Checklist for maintainers

• added entry in CHANGES.rst within the relevant release section
• updated or added relevant tests
• updated relevant documentation
• added relevant milestone
• added relevant label(s)
• ran regression tests, post a link to the Jenkins job below.
  How to run regression tests on a PR
• Make sure the JIRA ticket is resolved properly

[izkgao]

Would np.nansum be a better option?

import numpy as np
np.nansum([np.nan, 1.0, 2.0])
# 3.0
np.nansum([np.nan, np.nan, np.nan])
# 0.0