Annotation transcript choice - VariantAnnotation / VariantTranscriptAnnotation - canonical vs representative

At the moment we have VariantAnnotation which is a VariantTranscriptAnnotation plus a bunch of variant only fields

We populate VariantAnnotation (1 per variant) based on VEP pick (representative transcript) - this is what is used for filtering, and displaying on the analysis node grid.

It would be good to be able to filter across the transcript annotation - see #297

It would also be good to be able to choose which transcript to display, eg canonical transcript instead of representative

------

It may be worth seeing if we can break up VariantAnnotation to just have variant level stuff, then we join to the appropriate transcript data? The trouble is the transcript table is 5.6x bigger, then we have operations to decide what to display adding even more overhead.

Possibly - we could break up the 1-per transcript to be min/max for all transcripts and have the per transcript data into JSON fields? You may be able to just stop the filtering/column sorting