Faster exit for storage_stage client

[sakridge]

Problem

storage_stage can take a long time to exit if there is no leader. Causes test_replay to hang for a bit because the default timeout is 30 seconds.

Summary of Changes

Change the client timeout to 5 seconds and check exit on each iteration.

Fixes #

[codecov]

Codecov Report

Merging #2744 into master will increase coverage by <.1%.
The diff coverage is 76.9%.

@@           Coverage Diff            @@
##           master   #2744     +/-   ##
========================================
+ Coverage    79.1%   79.1%   +<.1%     
========================================
  Files         115     115             
  Lines       19174   19204     +30     
========================================
+ Hits        15169   15195     +26     
- Misses       4005    4009      +4

[sagar-solana]

Interesting that new_with_timeout exists. I feel like thin_client could use some refactoring to make sure all methods on thin_clients with timeouts, timeout. Now we've got some things that the rpc_client inside thin_client handles in terms of timing out and some things that fns within thin_client need to deal with (like the one I added).

[sakridge]

@sagar-solana yes I added it for another test case where it needed to timeout in a shorter amount of time.