I suggest a default value for fallbackValueType instead of making it optional; it does no harm and simplifies use – even if useFallbackValueAtEdge defaults to False (though shouldn't it default to True – especially since that's what the repair task wants?).

In that case the exception may be impossible to raise, so it's probably not worth trying to test it.

There is indeed a default for fallbackValueType and useFallbackValueAtEdge does default to True. I suppose this check is slight overkill, but it does no harm and prints an informative error message in the off chance the user specifies a discordant set of values for these config parameters.

As for the defaults, I tried to set them to what they effectively were in interpolateOnePlane (then run!) in interpImage.py. In repair.py, as you pointed out in your initial review, the default fallbackValue was being set differently, so I included the setDefaults to reflect what was being used there. I'm inclined to leave that as is since it is not unreasonable to want different settings for certain defect types (e.g. true defective CCD pixels versus a mask plane). This could certainly be revisited if the interpImage task becomes more frequently implemented and the callers are finding that a different set of defaults would be more appropriate.

Hey Lauren MacArthur - would you mind writing a (brief!) summary of this for https://confluence.lsstcorp.org/display/DM/Data+Release+Production+WIP+W16+release+notes?

Has been added to release notes – thanks!