Handling of non-positive values causing log10 warnings on EK calibration

[emiliom]

Handle values <= 0 (0 and negative values) in calibration compute to eliminate log10 warnings. For the dataarrays causing these warnings, sets non-negative values to a slightly positive non-zero value.

Addresses #737

Note: when I ran the test suite locally there were a bunch of errors unrelated to the warnings addressed here. I'm going to assume there's some other mess in my test setup or some temporary glitch, and let the CI tests take a crack. All the log10 warnings in test_range_integration.py are gone.

[emiliom]

There was just one failure in the CI, and it appears to be directly related to the change in this PR.

[leewujung]

@emiliom : Thanks! I will take a look tomorrow morning!

[leewujung]

I checked the CI failure and the code, and found the following:

• for the parts involving np.log10(tvg_mod_range): the warning will appear because the first small section of tvg_mod_range is negative, and the value will be NaN
• for the parts involving 10 * np.log10(prx): the warning will appear because the last element is 0, so the resulting value will be Inf

Let's change both so that the values are replaced with NaN before the log10 operation, since these the ranges that NaN at output does not impact practical functionality.

[emiliom]

Done. I've pushed a commit that replaces values with np.nan instead of 0. CI tests are churning.