Try to do less in-memory caching of compendiums

[ben]

Had a chat on the League discord not long ago, and it got me thinking. We're doing a lot of work to keep the compendiums fresh. I wonder if it might instead be better to do something like pack.getIndex({fields: ['system.dfid']}) to get index information, and then do async loads when we need actual documents.

[rsek]

already told ben about this a while back, but for posterity: i've experimented with this on local branches, and it does work. it can even point at flags, so we could index e.g. RollTables with flags.foundry-ironsworn.dfid