Bad performance leads to timeouts when writing to new shards

I've seen two community reports of bad write performance in 2.0.8:
* https://community.influxdata.com/t/influx-write-fails-with-internal-error
* https://community.influxdata.com/t/timeouts-severe-loading-data

Both are using `influx write` which should handle chunking large inputs into manageable batches, so I don't think either is sending more data to the server than it should be able to handle. Despite that, both show engine-level timeouts during write requests.

This could be an issue since 2.0 GA, but I'm suspicious of a regression since we're only starting to hear about it now.

Start by reproducing the engine-level write timeouts through `influx write`, using the two linked community reports as symptom references and focusing on writes to new shards. Compare behavior with the reported 2.0.8 cases and establish a repeatable test that confirms writes complete without timeouts.

Written by the indexing model from the issue text.