track fec set turbine stats

[jbiseda]

Problem

need to gather data on number of shreds received over turbine per fec set

(previous PR #23725)

Summary of Changes

• track shred count per fec set per slot in a per slot cache
• report data when entries are evicted from the cache

Fixes #

[codecov]

Codecov Report

Merging #23989 (4a516cf) into master (2da4e3e) will decrease coverage by 0.0%.
The diff coverage is 73.4%.

❗ Current head 4a516cf differs from pull request most recent head 1ea186d. Consider uploading reports for the commit 1ea186d to get more accurate results

@@            Coverage Diff            @@
##           master   #23989     +/-   ##
=========================================
- Coverage    81.8%    81.7%   -0.1%     
=========================================
  Files         581      585      +4     
  Lines      158312   159354   +1042     
=========================================
+ Hits       129518   130278    +760     
- Misses      28794    29076    +282     

[behzadnouri]

This commit still has several design issues; at least some were already discussed before: #23872 (comment)
You are also removing some of the constructs just added in #23872 which meant to simplify things, and I don't think that was the right thing to do.
But there has been too many back and forth on this metrics work, and I think it is better that you go ahead and merge this commit, and I will follow up with some patches to address the outstanding issues here.

[jbiseda]

This commit still has several design issues; at least some were already discussed before: #23872 (comment) You are also removing some of the constructs just added in #23872 which meant to simplify things, and I don't think that was the right thing to do. But there has been too many back and forth on this metrics work, and I think it is better that you go ahead and merge this commit, and I will follow up with some patches to address the outstanding issues here.

I've changed this to only report stats as entries are evicted from the cache. @behzadnouri I'll let you follow up if you think additional changes are required here. per our discussion, we're not going to get perfect data with a bounded cache. these metrics are for sanity checks.