Better handling of invalid cellIndices values

[psavery]

The cellIndices() function is used in the interpolation functions that we use for generating the polar view (for example, here).

However, sometimes nan values are produced within the function (as demonstrated in #729). These nan values, however, get converted to valid index values (like -1 or 0).

We should probably make sure we don't use invalid index values at all, rather than converting them to valid values.

Our best guess is that this bad handling of invalid cellIndices() values is resulting in bad interpolation values near detector borders. This is probably something that should be corrected.

[psavery]

We should set up some tests to see what this is actually doing, and how it is affecting results.

[psavery]

I tried looking into this. However, I could not reproduce the nan values. I'm not sure how they showed up. We should determine how they were showing up, and what we should do to fix them.