Suggested improvements for Metamist analysis IO

[EddieLF]

CPG Flow could use some changes to more cleanly interact with Metamist analysis records, first during workflow startup, and also at the point where analyses are written for the successfully completed stages.

Input

1. Querying for all cram and gvcf analyses upon startup
   I can see the merit to collecting all cram and gvcf paths, however it's not necessary for every workflow. For example, long-read cohorts or RNAseq cohorts do not need to check for gvcf existence. Done here during _populate_analysis, which runs for every workflow.

2. Async queries
   We should prefer to use asynchronous queries to Metamist

3. Using enums
   Addressing Analysis types in metamist.py not consistent with Metamist #128, CPG flow should query the Metamist enum tables for the allowed values, and then validate against these values from the enum table. See here for an implementation that queries the enum table and returns allowed types for each enum, not just the Analysis types.

Output

1. Allow writing of multiple analysis outputs to a single record. See Allow writing multiple files into a single analysis record #126