Investigate/Remove the limit at about 6KTPS of the native token benchmark w/ direct injection

[ssavenko-near]

Description

Running the benchmarking tool at benchmarks/transactions-generator with increasing target TPS does result in increase in accepted and processed TPS up to ~6KTPS(on my private PC). Any further increase in target TPS does not seem to contribute significantly to the actual TPS.
The 'unlimiting" changes to config are applied (by just do-it <TPS>).
The transactions do not even seem to be generated (see the logs below).
The PC is capable of generating about 30000 TPS at least in a single thread, so that should not be limiting factor.

Steps to reproduce

• run the tool in the benchmarks/transactions-generator with just do-it 5000
• run the tool with just do-it 50000
• observe the actual transaction rate being roughly the same

Logs extracts

$ just do-it 5000
...
2025-02-17T09:34:06.673225Z INFO transaction-generator: diff="Stats { pool_accepted: 5764, pool_rejected: 0, processed: 6450, failed: 0 }"
2025-02-17T09:34:07.455162Z INFO transaction-generator: diff="Stats { pool_accepted: 4884, pool_rejected: 0, processed: 3401, failed: 0 }"
2025-02-17T09:34:08.565412Z INFO transaction-generator: diff="Stats { pool_accepted: 4767, pool_rejected: 0, processed: 6607, failed: 0 }"
2025-02-17T09:34:09.455274Z INFO transaction-generator: diff="Stats { pool_accepted: 5201, pool_rejected: 0, processed: 3050, failed: 0 }"
...

$ just do-it 50000
2025-02-17T09:36:16.258822Z INFO transaction-generator: diff="Stats { pool_accepted: 6265, pool_rejected: 0, processed: 6441, failed: 0 }"
2025-02-17T09:36:17.055366Z INFO transaction-generator: diff="Stats { pool_accepted: 4952, pool_rejected: 0, processed: 4104, failed: 0 }"
2025-02-17T09:36:18.113895Z INFO transaction-generator: diff="Stats { pool_accepted: 4855, pool_rejected: 0, processed: 6501, failed: 0 }"
2025-02-17T09:36:19.055316Z INFO transaction-generator: diff="Stats { pool_accepted: 6141, pool_rejec