workup methods should do preliminary check to ensure channel of interest is not exclusively 0 or nan

[darienmorrow]

This is merely a convenience feature, but it is a nice check.
@jscheeler1 was very confused because his splines were failing, but the base problem was actually an all nan channel.

[darienmorrow]

A better solution is to check that the channel is not exclusively composed of a single value. I think one way to check this would be to raise an error if np.unique(chan[:]).size > 1 is not true.