Bulk CSV: Upload 10k cases in under 1 minute

[axmb]

Describe the solution you'd like
We want to establish a reasonable upper bound on the recommended data size for bulk upload. We've chosen that to be 10k cases, and since it's advertised as an acceptable limit, we want it to complete in a similarly reasonable amount of time. For that, we've chosen <1 minute.

Describe alternatives you've considered

• Bulk upload is slower, but provide progress (%) feedback (would rather have the significantly better speed).
• Set a lower upper bound (infeasible; need to support at least ~10k).
• Target adding more cases in the same time; e.g., an upper bound of 50k (will instead suggest this type of upload be handled via automated ingestion).

Additional context
This is also relevant to automated ingestion, which will use the same APIs as bulk upload. Automated ingestion, however, has an easier time sending large amounts of data. Since it's sent from a Docker image running with O(GB) of memory, we don't have to dance around sharding data in and out of the browser.

[axmb]

Related PRs:

#584
#594
#605
#642
#680
#691
#701

[axmb]

I'll retest this shortly. We were close to 10k/minute, but added some slowdowns along with keeping case revisions (which were unfortunately necessary). The slowdowns do impact ADI, but since ADI functions have 15 minutes to complete and don't require a human's attention, I'll leave this at P2 unless that 15 minute deadline is in jeopardy.

[axmb]

Circling back -- there were some issues around this, but we've mostly resolved them.

There's a separate issue tracking a related FR in ADI (#1124).