Input a bad channel list (3D tools)

[Cameron-Van-Eck]

From Paddy Leahy:

Wishlist item for RM-Tools: a "bad channel" list. At the moment you can simulate this by giving it a noise file with very large noise values for channels you want to exclude, but explicitly ignoring them would be neater and also probably save a bit of comp time.
One application is to check for impact of finite short-spacing coverage on diffuse emission, by cutting out the top part of the band - on the basis that RM gradients will chop up Q,U in the bottom part of the band more effectively.

My thoughts:
This can nominally be skipped for the 1D tools -- it should be minimal fuss for a user to edit the input text file to remove any channels they wish. For the 3D tools, where it's more fuss to blank out channels, this could be useful. How best to do this? Add a new input: either a string with channel selections, or a text file? Could this be done in a clever way using the existing noise file methods -- perhaps by allowing channel noise values to be NaNs, which would then indicate a bad channel? (Maybe that's even already supported -- I'd have to check). The slightly fussy thing would be to make it work even when the user has specified uniform weighting -- it currently completely ignores the supplied noise vector in this case.