1arcsec imaging fixes

[jurjen93]

This MR fixes a few issues for the 1" imaging workflow:

• Provide number of CPU as input parameter for the average_intermediate_resolution.cwl step. This is tied to the ncpu for imaging, which is handy if the user runs the entire workflow on a single node (which will be probably the preferred method on clusters such as Spider).
• Give more default number of cores for the 1" imaging, as 12 cores was too few.
• A bug where the imaging took the input MS of the workflow as input, instead of the BDA-averaged output from the first step
• Allowing the user to give a tmpdir directory for wsclean imaging
• A bug for returning the wrong non-pb images after imaging