regrid clean mask cube to broader channel width fails

[Fil8]

the line_worker fails when regridding the mask to clean the datacube.

The user may want to create a datacube with a different binsize than the original dataset (the input .ms file). When this happens, sofia creates a cleanmask with naxis3 of same size of the requested output datacube.

Then the pipeline crashes when checking the size of the cleanmask, right before cleaning.

# 2024-01-06 02:19:34: Running mProjectCube /stimela_mount/input/NGC3557_r10_t90_3kms_3kms_cleanmask_0_mask_01.fits /stimela_mount/output/NGC3557_r10_t90_3kms_3kms_cleanmask_
0_mask_comb_NGC3557_regrid.fits /stimela_mount/input/tmp.hdr
# [struct stat="OK", module="mProjectCube", time=5.0]
# INFO:    Converting SIF file to temporary sandbox...
# INFO:    Cleaning up image...
2024-01-06 02:19:40 CARACal.Stimela.reprojectMask-field0 INFO: job complete at 2024-01-06 02:19:40.344020 after 0:00:15.385101
2024-01-06 02:19:40 CARACal.Stimela.line__comb INFO: Saving pipeline information in .last_line__comb.json
2024-01-06 02:19:40 CARACal.Stimela.line__comb INFO: Recipe executed successfully
2024-01-06 02:19:40 CARACal INFO: Reprojecting mask NGC3557_r10_t90_3kms_3kms_cleanmask_0_mask_comb_NGC3557_regrid.fits to match the spectral axis of the cube.
2024-01-06 02:19:40 CARACal ERROR: Requested channels are not contained in mask NGC3557_r10_t90_3kms_3kms_cleanmask_0_mask_comb_NGC3557_regrid.fits. [OSError]

[Fil8]

I suspect it is because in this case the pipeline enters in this loop, because it does not enter in any other ifs.

caracal/caracal/workers/line_worker.py

Line 981 in 5bafd40

doSpec = None # likely will fail/produce incorrect result in the case that the ms file and mask were not created with the same original spectral grid.

[Fil8]

or maybe is because of this typo:

caracal/caracal/workers/line_worker.py

Line 968 in 5bafd40

dpSpec = None # this should work in both a request for a subset, and if the cube is to be binned.

[Fil8]

I also found out that the pipeline fails if a binned mask is given as input.

For example one can make cube_0 binning channels, run sofia to create a mask and then make cube_1. At this point line_worker.py fails because it counter checks the size of the mask against the size of the ms-file and product for binchans is forgotten at this line.

caracal/caracal/workers/line_worker.py

Line 979 in 2d338e2

if np.round(cdelt,1) > np.round(chanwidth[0],1):

this is now solved in maskForn repo:

caracal/caracal/workers/line_worker.py

Line 960 in f5aeb81

if np.round(cdelt) > np.round(chanwidth[0]*binchans,5):

The issue above still needs to be solved.