Closed rhiannonlynne closed 1 year ago
The scope of this PR expanded once I added a unit test to make sure that the SaturationStacker ran. I set up the unit test to run all of the possible stackers (i.e. not BaseStackers or moving object stackers, purely because I didn't have an easy data source for those .. but it would be good to add next). This meant fixing up a couple of other stackers which didn't work with a full datasource automatically or maybe worked a little less ideally than we would typically use them now. I then tossed all of the dither stackers, as many of them required "fieldID" which is not a stacker we provide any longer. This then meant removing dither stacker references from other locations (in the batches, primarily).
If load_inst_zeropoints is in init method, then the actual throughput files are attempted to be read as soon as the stacker is instantiated (as in the batches or sometimes in metrics init methods). Moving the call to load_inst_zeropoints to the
run
method lets rubin_sim import even without data being available, instead of failing to import in that case.