Continuum pipeline flow - round1 waits for all beams

[tjgalvin]

In the current continuum imaging and self-calibration workflow there is a draining of works waiting for all beams to finish imaging + source finding while the initial set of images (pre self cal) are produced. This means that some significant mount of time may be wasted waiting for one beams to finish image. The other beams should still be able to go through all rounds of self-cal.

I believe that this is happening for two potential reasons:

• Getting the MSSummary object after imaging. If this is not respected properly the self-cal task would zip the MS up, and the MSSummary would be unable to run.
• waiting for all images to complete to get a common beam out, which might go into the FieldSummary

In either cases, one of these two stages is getting itself into the dependencies of the self-calibration task. I see that the masking stages all happening asap, so there is something going into the self-cal task that is not in the masking tasl that is causing things to block.

[tjgalvin]

Perhaps the better approach is to make all the images in on go, and only bother with the linmos and validation plots at the very end. This is alternative to the current form, which is image -> linmos -> image -> linmos.

By linmos I mean convolving to common beam, co-adding, source finding.