Pull request #1073: CAS-14738
Merge in CASA/casa6 from CAS-14738 to master
* commit '34e7f8d722235aea9f1e51000bda6e42287399d7':
indicate in the XML that use_durations is experimental, add final planned checks to checkActualAxes
move some shared code into separate functions to avoid code duplication.
propogate actualDurations warnings up to the the sdm tool for logging
moved some use_durations tests from the sdm tool test_task_importasdm
use_durations added at the top, test code added
values set in VMSData structure to be transferred to MS CAS-14738 : actual* fields are used where appropriate, still rough around the edges and not used out of the lower level, but worth committing finally.
Pull request #1071: CAS-14778
Merge in CASA/casa6 from CAS-14778 to master
* commit '0d731f5701553c3bcd07344a0be600f6ec4afee4': CAS-14778 : Fix rectangular image case
rename to cameltype and assert that dimensions are either both even or both odd
Add test
Copy the solution for odd pixels that was found for Matrix Cleaner over to also apply to MultiTermMatrixCleaner.cc Validated on Cygnus A, scales=[0,1], nterms=1, cellsize = 1.6 arcsec (such that the bright termination lobes correspond to just a few pixels transversely, in this way we can test the effect of the code change without simply looking at the effects of extended emission)
Solved the problem for an odd number of pixels for Asp-CLEAN and MS-CLEAN. That is simply done by moving the slice for the beam subtraction up by one pixel. The problem does not exist for Hogbom. TBD: Verify that these fixes also fix the spectral versions and multiterm versions of the deconvolvers.
