Handling of Mosaics in MSv4

[dpetry]

As discussed with Jan-Willem, the MSv4 originally was meant to contain data for only one Field
such that multi-field MSs would need to be stored in several MSv4 inside one Processing Set.
Later the MSv4 was modified to permit more than one Field.
So, for mosaics, i.e. observations of the same target with the same array in one execution block
sequentially moving over a set of Fields (in the ASDM and MSv2, they all have the same Field Name
but not the same Field ID), the questions are:

1. Do the data belonging to the many fields (potentially hundreds) of the same mosaic all get stored
   in one MSv4 or do they get spread over many MSv4s?
   What should be the default behaviour?
2. How can a piece of software reading the data recognise that it is dealing with a Mosaic?
3. How can the reading software determine the extend of the Mosaic (total number of fields, spatial extent)?
   In other words, how is the metadata of the mosaic set in the first place and then made available?