Add Niagara system and Planck NPIPE

[msyriac]

Not adding Penn GPC for now.

[msyriac]

Note that for Planck NPIPE, the split numbers are 1 and 2, rather than 0 and 1. And the map tags are slightly different. But sofind seems flexible enough not to care.

[zatkins2]

This is all great, thanks for adding.

For Planck, that is good to know. You're right that sofind won't care, but mnms will, because it does iteration over splits in couple places like range(num_splits). simonsobs/mnms#10

Could you also target develop for this PR?

[msyriac]

How do you go about addressing the split number filename mismatch? (I'll work on targeting to develop)

[zatkins2]

I don't really see a way around cluing mnms to expect some other iterable of splits. So somehow that information would need to be recorded explicitly (e.g., if not added, assume 0-indexed, otherwise, some item in a yaml that just lists out the split iterable). This would likely be an addition to these Planck yamls you just added, but let's worry about that when we know what we want.