[attraction] reduce allocations in HASH action

[constanca-m]

Description

Improves performance on hashing. See:

goos: linux
goarch: amd64
pkg: github.com/open-telemetry/opentelemetry-collector-contrib/internal/coreinternal/attraction
cpu: 11th Gen Intel(R) Core(TM) i7-11800H @ 2.30GHz
                                │ current code │               this PR                │
                                │    sec/op    │    sec/op     vs base                │
AttrProc_HashRegex/N20_R10-16      2.578µ ± 1%   2.354µ ± 14%   -8.69% (p=0.000 n=10)
AttrProc_HashRegex/N20_R50-16      5.763µ ± 2%   4.497µ ±  6%  -21.97% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R10-16    191.7µ ± 5%   145.8µ ±  2%  -23.98% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R50-16    530.6µ ± 6%   409.3µ ± 12%  -22.87% (p=0.000 n=10)
geomean                            35.06µ        28.19µ        -19.60%

                                │ current code  │               this PR                │
                                │     B/op      │     B/op      vs base                │
AttrProc_HashRegex/N20_R10-16        353.0 ± 0%     161.0 ± 0%  -54.39% (p=0.000 n=10)
AttrProc_HashRegex/N20_R50-16       1762.0 ± 0%     801.0 ± 0%  -54.54% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R10-16    34.47Ki ± 0%   15.70Ki ± 0%  -54.45% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R50-16   172.08Ki ± 0%   78.25Ki ± 0%  -54.53% (p=0.000 n=10)
geomean                            7.702Ki        3.506Ki       -54.48%

                                │ current code │               this PR               │
                                │  allocs/op   │  allocs/op   vs base                │
AttrProc_HashRegex/N20_R10-16       8.000 ± 0%    4.000 ± 0%  -50.00% (p=0.000 n=10)
AttrProc_HashRegex/N20_R50-16       40.00 ± 0%    20.00 ± 0%  -50.00% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R10-16     800.0 ± 0%    400.0 ± 0%  -50.00% (p=0.000 n=10)
AttrProc_HashRegex/N2000_R50-16    4.000k ± 0%   2.000k ± 0%  -50.00% (p=0.000 n=10)
geomean                             178.9         89.44       -50.00%

Testing

I added a unit test to make sure there were no breaking changes.

[iblancasa]

Shouldn't this be something like this?

  template := newMap(b, tc.totalAttrs, tc.secretRatio)

  b.ReportAllocs()
  for b.Loop() {
      attrs := pcommon.NewMap()
      template.CopyTo(attrs)
      ap.Process(b.Context(), nil, attrs)
  }

I guess the loop is iterating over the same attrs, so we are just doing real work on the first iteration.

[constanca-m]

Yes 🤦 I have updated the issue description with the new results. All good now?

[constanca-m]

Hey @iblancasa , can you take a look at this again? Also @codeboten since you are assigned

[atoulme]

LGTM