Bug: TH warnings

[nohelix]

thresholding (i.e. psycho) likes to spit out warnings in different cases

Example bad files found in Z:\Issues:

• 9/26/22 LP3 Note this also gives an additional error of unknown column FA
• 8/29/22 BP3

[nohelix]

So the warnings are typically do to low trial counts and trying to fit a line to a weak model. It seems like we should try to capture the warnings from approx() that are returned during the map and convert them to a single warning in warnings_list that says "Calculated TH is questionable to to poor modeling conditions."

Not sure how to program that.