add option for resource attributes in metrics for prometheus exporter

[codeboten]

This PR adds the WithResourceAsConstantLabels option to the Prometheus exporter to allow users to configure resource attributes to be applied on every metric.

Fixes #4732

[codecov]

Codecov Report

Merging #4733 (b7cb11c) into main (0405492) will increase coverage by 0.0%.
The diff coverage is 90.6%.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #4733   +/-   ##
=====================================
  Coverage   81.9%   81.9%           
=====================================
  Files        224     224           
  Lines      18116   18136   +20     
=====================================
+ Hits       14847   14865   +18     
- Misses      2982    2984    +2     
  Partials     287     287           

Files	Coverage Δ
exporters/prometheus/config.go	100.0% <100.0%> (ø)
exporters/prometheus/exporter.go	84.7% <89.4%> (+0.7%)	⬆️

... and 1 file with indirect coverage changes

[MadVikingGod]

It looks good so far.

[MadVikingGod]

With there being minimal change to the no-filter path I think this is gtg.

There is a cost to add these attributes in the filter path. Adding a service.name filter I got these results:

benchstat bench-no-filter.txt bench-with-filter.txt
name             old time/op    new time/op    delta
Collect1-16        32.3µs ± 5%    33.4µs ± 3%   +3.45%  (p=0.002 n=10+10)
Collect10-16       93.2µs ± 3%   101.7µs ± 2%   +9.04%  (p=0.000 n=11+9)
Collect100-16       529µs ± 1%     613µs ± 4%  +15.95%  (p=0.000 n=9+10)
Collect1000-16     4.97ms ± 2%    5.64ms ± 5%  +13.63%  (p=0.000 n=11+10)
Collect10000-16    43.4ms ± 5%    50.3ms ± 7%  +15.83%  (p=0.000 n=11+9)

name             old alloc/op   new alloc/op   delta
Collect1-16        35.5kB ± 0%    35.8kB ± 0%   +0.74%  (p=0.000 n=11+10)
Collect10-16       48.4kB ± 0%    51.1kB ± 0%   +5.45%  (p=0.000 n=11+10)
Collect100-16       193kB ± 0%     219kB ± 0%  +13.70%  (p=0.000 n=11+10)
Collect1000-16     1.57MB ± 0%    1.83MB ± 0%  +16.85%  (p=0.000 n=10+10)
Collect10000-16    15.1MB ± 0%    17.8MB ± 0%  +17.57%  (p=0.000 n=11+10)

name             old allocs/op  new allocs/op  delta
Collect1-16          72.0 ± 0%      78.0 ± 0%   +8.33%  (p=0.000 n=11+10)
Collect10-16          407 ± 0%       467 ± 0%  +14.74%  (p=0.000 n=11+10)
Collect100-16       3.76k ± 0%     4.36k ± 0%  +15.97%  (p=0.000 n=11+10)
Collect1000-16      37.1k ± 0%     43.1k ± 0%  +16.16%  (p=0.000 n=11+10)
Collect10000-16      372k ± 0%      432k ± 0%  +16.20%  (p=0.000 n=11+10)

[dashpole]

@MadVikingGod note that it is benchmarking only the first scrape of an endpoint (a single Collect for the exporter). Since the result is cached, it should have minimal impact on subsequent scrapes.