Allow extrapolating a little bit more in frequency

Fixes lofar-astron/dp3#326

This accommodates an edge case where solutions have been produced on an averaged data set, where the edge channels were ignored. This can occur when the total number of channels on the calibrator was not divisible by the averaging factor, e.g. total number of channels is 488 and averaged by a factor 10, then the first and last 4 channels can be ignored. With this fix, these solutions on an averaged data set can still be applied to the target on all channels.

approved this merge request

merged

mentioned in commit 1d38da16

mentioned in merge request !48 (merged)