Release scalable continuum imaging pipeline

Continuum imaging is a critical part of both self-calibration and the generation of continuum imaging cubes. It is also the hardest pipeline to scale effectively, as it may need to deconvolve across very broad bands (as well as calibrate, in case of self-calibration).

Inputs

1. Final DD calibration solutions from self-calibration
2. Averaged/calibrated visibilities
3. Local sky model

Outputs:

1. Continuum imaging cube data product
2. Source catalogue data product

Release and demonstrate visibility streaming pipeline

1. Capable of reading visibility dataset of significant size and perform multiple major cycles (i.e. including deconvolution)
2. Show distribution using Dask, and reason about overall scalability
3. Demonstrate that we can generate realistic images