Decide where the CFA files for CANARI are going

[bnlawrence]

The cache on JASMIN. Presuming these will be a GWS initially, but could go to object store when we have netCDF S3.

CFStore will be creating CFA files that it will need to see for ingestion, both at the original cycle and when re-aggregating as cycles are added. (So the aggregation will likely see the "existing aggregation CFA" and the next cycle, after which it will discard the next aggregation in favour of the new grand aggregation.)

WE also anticipate that users may want to use cf-python directly on these aggregation files.