Comparison of float bug in radiance_irs.py

[karpob]

fix.txt

When trying to duplicate the largest pc score behavior in the ioda converter I've been developing, I noticed the comparison of two float values with an ==. There are two problems with this:

1. It's bad practice and is likely to create bugs down the line depending on numpy updates, etc.
2. It is impossible to create a vectorized version of this as the == has irreproducible behavior.

I'm still working on implementing the vectorized version in my code and you may want to look at it, but short-term I'd recommend the following patch (sorry, can't create a branch/pull request no permissions to do so).