Make models download complicated attributes on-demand

[MegaApuTurkUltra]

So by default Resource shouldn't download owners, and Collection shouldn't download items, owners, curators, or subscribers, since those could potentially get quite large. Instead, checking if an ID is on that list should be done as a mongo query and all actual downloads of the list should be paginated. This kind of goes with #41

[bates64]

https://docs.mongodb.com/manual/reference/operator/query/nin/#op._S_nin

[MegaApuTurkUltra]

Partially implemented in Collection 🎉

[MegaApuTurkUltra]

Somewhat complete. Resources.downloaders could still use some work

[MegaApuTurkUltra]

Just did Resource.downloaders. The last things should be collection owners/curators/subscribers (although I think owners should be ok - we don't expect a huge number of owners anyway - and curators/subscribers don't have any non-mongo-based operations so far 📦 ), and comments, where we probably want them paginated