Make active series benchmark more realistic, improve performance of context cancellation check.

[flxbk]

What this PR does

This makes the responses used in BenchmarkActiveSeriesMiddlewareMergeResponses more realistic.
Using the altered benchmark, it improves performance of the context cancellation check performed in mergeResponses.

pkg: github.com/grafana/mimir/pkg/frontend/querymiddleware
                                                                             │   main      │              branch               │
                                                                             │   sec/op    │   sec/op     vs base              │
ActiveSeriesMiddlewareMergeResponses/seriesCount=10_000/num-responses-4-10     19.05m ± 9%   19.14m ± 1%       ~ (p=0.818 n=6)
ActiveSeriesMiddlewareMergeResponses/seriesCount=10_000/num-responses-16-10    90.16m ± 2%   86.81m ± 2%  -3.72% (p=0.002 n=6)
ActiveSeriesMiddlewareMergeResponses/seriesCount=10_000/num-responses-64-10    422.4m ± 4%   401.8m ± 6%  -4.88% (p=0.002 n=6)
ActiveSeriesMiddlewareMergeResponses/seriesCount=10_000/num-responses-128-10   924.7m ± 3%   849.8m ± 6%  -8.10% (p=0.004 n=6)
geomean                                                                        160.9m        154.3m       -4.10%

Which issue(s) this PR fixes or relates to

Fixes #

Checklist

• Tests updated.
• Documentation added.
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX].
• about-versioning.md updated with experimental features.

[ortuman]

looks good! thanks! 👍