Improve concurrent performance of exponential histogram measurements

[dashpole]

This builds on #7474 and #7478. It is similar to the histogram implementation, but uses an additional hotColdWaitGroup for downscaling without blocking the hot path, and a "circular" slice for buckets to avoid shifts.

This is not ready for review, and still needs a significant amount of polish. I will return to this after the previous PRs are merged.

                                                                                  │   main.txt   │              hist.txt               │
                                                                                  │    sec/op    │    sec/op     vs base               │
SyncMeasure/NoView/ExemplarsDisabled/ExponentialInt64Histogram/Attributes/0-24      296.4n ± 18%   155.5n ±  4%  -47.54% (p=0.002 n=6)
SyncMeasure/NoView/ExemplarsDisabled/ExponentialInt64Histogram/Attributes/1-24      304.0n ± 16%   152.1n ±  2%  -49.97% (p=0.002 n=6)
SyncMeasure/NoView/ExemplarsDisabled/ExponentialInt64Histogram/Attributes/10-24     332.8n ±  5%   150.4n ±  4%  -54.80% (p=0.002 n=6)
SyncMeasure/NoView/ExemplarsDisabled/ExponentialFloat64Histogram/Attributes/0-24    314.7n ±  8%   152.6n ± 16%  -51.51% (p=0.002 n=6)
SyncMeasure/NoView/ExemplarsDisabled/ExponentialFloat64Histogram/Attributes/1-24    314.2n ±  8%   158.7n ±  3%  -49.49% (p=0.002 n=6)
SyncMeasure/NoView/ExemplarsDisabled/ExponentialFloat64Histogram/Attributes/10-24   334.3n ±  8%   157.2n ±  3%  -52.96% (p=0.002 n=6)
geomean                                                                             315.8n         154.4n        -51.10%

[codecov]

Codecov Report

❌ Patch coverage is 87.86127% with 42 lines in your changes missing coverage. Please review.
✅ Project coverage is 86.1%. Comparing base (f57bf14) to head (8871f03).

Files with missing lines	Patch %	Lines
sdk/metric/internal/aggregate/atomic.go	23.3%	23 Missing ⚠️
...metric/internal/aggregate/exponential_histogram.go	93.9%	11 Missing and 8 partials ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##            main   #7535     +/-   ##
=======================================
- Coverage   86.2%   86.1%   -0.1%     
=======================================
  Files        298     298             
  Lines      21829   22003    +174     
=======================================
+ Hits       18817   18957    +140     
- Misses      2635    2664     +29     
- Partials     377     382      +5     

Files with missing lines	Coverage Δ
sdk/metric/internal/aggregate/aggregate.go	100.0% <100.0%> (ø)
...metric/internal/aggregate/exponential_histogram.go	94.8% <93.9%> (-2.6%)	⬇️
sdk/metric/internal/aggregate/atomic.go	75.7% <23.3%> (-12.5%)	⬇️

... and 3 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[dashpole]

I'm going to try to split this into smaller PRs:

• Migrate to circular buffer to avoid shifts (which we can't do in a lockless manner)
• Compute count at export time
• Split scale by positive and negative, and re-join at collect time.