BUG - exclusion of zeros in SWAPI I-ALiRT fitting

[hafarooki]

Description of the issue

For times when there is a zero count in one of the passbands near the proton peak in SWAPI, which is more common than expected because the I ALiRT counts are rounded down, the initial guess is returned as the fit by the algorithm. Need to exclude zeros from the fit. Extracted from #2485 to be fixed in a new PR because the PR scope had grown too broad.

Steps to reproduce the issue

No response

Expected behavior (What should happen)

No response

Actual behavior (What does happen)

No response

Code Snippet:

Code

Additional notes, affected areas, and suggested fixes

No response