metrics: add counters to broadcastSignedTxGroup

[cce]

Summary

Signed transactions intended for broadcast from a node (via the REST API) go through this broadcastSignedTxGroup method for validation first. This adds counters to track how often they succeed vs fail.

Test Plan

Existing tests should pass.

[codecov]

Codecov Report

Merging #5588 (394adc1) into master (d316914) will decrease coverage by 0.01%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #5588      +/-   ##
==========================================
- Coverage   55.79%   55.78%   -0.01%     
==========================================
  Files         446      446              
  Lines       63418    63423       +5     
==========================================
- Hits        35381    35378       -3     
- Misses      25668    25673       +5     
- Partials     2369     2372       +3     

Impacted Files	Coverage Δ
node/node.go	4.16% <0.00%> (-0.03%)	⬇️

... and 5 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[iansuvak]

Generally LGTM, the only comment is around the naming of the metrics. The PR explains that the benefit of these is that we get to count successes/failures of broadcasting messages received via REST API specifically but nothing else about their names or about the method suggests that this codepath is only used for txns received via REST.

Not a blocker either way since people can trace through the code and see it or find the PR but just digging through a list of /metrics might be non-intuitive.

[algorandskiy]

Not a blocker either way since people can trace through the code and see it or find the PR but just digging through a list of /metrics might be non-intuitive.

I agree, metric names have to be re-unified. Maybe make a cleanup by introducing new names and then retiring the old ones in next release