Perhaps refactor the WCU_OFF processing into a separate recipe?

[hugobuddel]

Some calibration templates (e.g. METIS_img_lm_cal_DetLin) produce their own dark-like exposures, which are currently processed in the same recipe as the main calibration data of those templates.

However, it seems more sensible to just process those WCU_OFF darks with the metis_det_dark recipe and then use the resulting kinda-like-masterdark as input to the main recipe.

Put for after FDR, but wouldn't surprise me if we get RIXed for it.