Add get_blocks and get_slot methods to bench-tps-client

[KirillLykov]

Problem

For the analysis of the confirmed transactions implemented in #92 BenchTpsClient needs some additional methods.

Summary of Changes

Add get_block, get_blocks and get_slot methods.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 81.8%. Comparing base (85cfe23) to head (318a67b).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff            @@
##           master      #94     +/-   ##
=========================================
- Coverage    81.8%    81.8%   -0.1%     
=========================================
  Files         838      838             
  Lines      226223   226223             
=========================================
- Hits       185240   185177     -63     
- Misses      40983    41046     +63     

[bw-solana]

Plan is still for ThinClient to be deprecated then deleted, right?

[gregcusack]

Yes, been deprecated in PR: #35365. Fully removing it from other parts is WIP

[bw-solana]

Code looks good to me.

Design wise, I'm debating if bench-tps is the right tool for doing this. I can convince myself that verifying transaction results is a reasonable extension of the existing tool.

[ilya-bobyr]

Questions about the commitment level are quite reasonable.
This change is to support more metrics in bench-tps.
Specifically, Kirill is adding block fetching code here: #92
As bench-tps is supposed to be a major source of transactions, while it is running, fetching whole blocks and getting transaction stats this way it probably the most efficient approach.

There is a short summary here: https://github.com/anza-xyz/agave/pull/92/files#diff-1d0a43d64af632a7d6ecf76ff095dec3b9a6243758fb5818752e26f4c5383611R1-R2

But it would not hurt to provide a more extensive reasoning in the same or some other comment, relevant to the added service :P

For fetching blocks, I think, we want confirmed commitment level. Latency is not important, and we would not want to deal with forks.
We may want to add some fork related stats in there at a later point.

[KirillLykov]

Code looks good to me.

Design wise, I'm debating if bench-tps is the right tool for doing this. I can convince myself that verifying transaction results is a reasonable extension of the existing tool.

Alternatively, I could have implemented a standalone tool which goes over all the provided signatures in the give span of blocks and does the same. This approach will make bench-tps runs execution to take much longer time and will lead to slightly more complex setup (create signature file, read it, what if something went wrong, etc)

[bw-solana]

LGTM