Saturate multiple shards with the Locust FT transfer benchmark

[aborg-dev]

We can saturate a single shard with 550 TPS, but two shards also peak at around the same number which is not expected - the throughput should be roughly double that. We should investigate why this is not the case and fix it.

[MCJOHN974]

Upd: document with instructions how to saturate two shard network (and get around 800 TPS on a bit weaker machine comparing to one that produces 550 TPS on single shard)

TODO: refactor the document, probably create a new one with this instruction only.

[MCJOHN974]

@Akashin, probably solution "just use multiple locusts or multi-machine locust" is good enough for our purposes and we can close this?