Add new metadata to UVCal and UVData objects to track the polarization normalization

[bhazelton]

There are two potential approaches:

• Add a new optional UVParameter to the objects, maybe called polarization_normalization, to track this. The options should be "sum" and "average" to indicate whether the two polarizations should be averaged or summed to get to Stokes I. It shouldn't be set for uncalibrated data or for pseudo-stokes polarizations (unless you're doing something very strange).

• Fold this into the vis_units parameter, by indicating 'Jy (sum)' or Jy (average)'. Jy would still be allowed.

Note: we'd only do one of the above options, interested in community feedback about which one to implement.

When applying a calibration to a UVData object, if this metadata is set on the UVCal object it should be propagated to the calibrated UVData object.

[jsdillon]

This seems reasonable to me.

[bhazelton]

This seems reasonable to me.

@jsdillon Do you have a preference between the two options?

[jsdillon]

Sorry, I was confused. I thought the idea was to do both.

I think Option A feels more "right" to me, but I'm open to both. I don't feel like you have to do both.

[rlbyrne]

I prefer the first option, to keep vis_units a real unit. I think the second option could also get confusing because there is nothing in the parameter or value to connect "sum" and "average" to polarization.

[jsdillon]

We're looking to change over the HERA delay spectrum pipeline to the "sum" convetion and it'd be helpful to have this new optional parameter (which seems like the consensus) in place when we do that. Would it be possible to get this added this week?