refactor(share/shrex-nd): blacklist peers that send invalid data

[walldiss]

Overview

Shrex should blacklist peers that send bad data.

[codecov-commenter]

Codecov Report

Merging #2231 (2db5afe) into main (e8ffd64) will increase coverage by 0.11%.
The diff coverage is 42.10%.

@@            Coverage Diff             @@
##             main    #2231      +/-   ##
==========================================
+ Coverage   55.81%   55.92%   +0.11%     
==========================================
  Files         216      216              
  Lines       14130    14141      +11     
==========================================
+ Hits         7886     7909      +23     
+ Misses       5454     5442      -12     
  Partials      790      790              

Impacted Files	Coverage Δ
share/availability/discovery/discovery.go	73.85% <0.00%> (-2.40%)	⬇️
share/getters/shrex.go	77.12% <0.00%> (-1.55%)	⬇️
share/p2p/shrexnd/client.go	69.42% <ø> (+1.42%)	⬆️
share/availability/discovery/metrics.go	47.05% <57.14%> (+0.63%)	⬆️
share/availability/discovery/backoff.go	86.79% <100.00%> (+1.07%)	⬆️

... and 3 files with indirect coverage changes

[renaynay]

this is also very good because now, if we have malicious peer who sends us bad data, it won't short-circuit the request, it'll just blacklist and keep going until it gets good shares. However, there's an issue with the err inside of GetSharesByNamespace in shrex getter: it's only ever returned in the case of not being able to get a peer from peer manager, but not in the case of a context timeout. Why is this the case? Shouldn't we errors.Join the context err as well during timeout and return it? esp since we doing have any checks for ctx deadline exceeded in cascade getter

Also this would be a very good candidate for swamp test.

[walldiss]

However, there's an issue with the err inside of GetSharesByNamespace in shrex getter: it's only ever returned in the case of not being able to get a peer from peer manager, but not in the case of a context timeout. Why is this the case? Shouldn't we errors.Join the context err as well during timeout and return it? esp since we doing have any checks for ctx deadline exceeded in cascade getter

Nice find, we used to have all errors joined before relatively recent hotfix this one I'll open another PR to fix it, since it is unrelated to this change and there is also same problem in GetEDS

[renaynay]

@walldiss needs rebase