flatCorrect() [and other primitives?] do some checks on user-supplied calibrations

[chris-simpson]

The imaging version of flatCorrect() checks that the flat has the same filter as the science. It should not do this as the user should be able to submit any file as a flatfield (and if the calibration comes from the caldb then the filter will match anyway). We should prevent such failures (although it may be wise to log a WARNING if there is a filter mismatch).

The other primitives that similarly apply a calibration should be checked also.