Implement WithExplicitBucketBoundaries option in the metric SDK

[dashpole]

Fixes #4094

I looked at both #4341 and #4340 when implementing this, but mostly relied on #4340. I omitted changes to the existing lookup() and aggs() functions (made in #4340) to minimize the diff. If we still want those changes, I'd prefer refactoring as a follow-up.

The precedence of bucket boundaries is implemented as follows:

1. Boundaries specified in the View
2. Boundaries specified on the instrument
3. Boundaries provided by the reader, or the default boundaries if the reader uses the DefaultAggregation

[codecov]

Codecov Report

Merging #4605 (5d0cea9) into main (5ec67e8) will increase coverage by 0.0%.
The diff coverage is 88.0%.

@@          Coverage Diff          @@
##            main   #4605   +/-   ##
=====================================
  Coverage   81.6%   81.7%           
=====================================
  Files        225     225           
  Lines      17991   18047   +56     
=====================================
+ Hits       14692   14753   +61     
+ Misses      2999    2996    -3     
+ Partials     300     298    -2     

Files	Coverage Δ
sdk/metric/meter.go	89.0% <100.0%> (+3.1%)	⬆️
sdk/metric/pipeline.go	86.3% <77.2%> (+0.4%)	⬆️

[MadVikingGod]

I'm ok with this overall.

How much do you think would need to change if there was a WithExponential() option?

[dashpole]

How much do you think would need to change if there was a WithExponential() option?

You would basically just change HistogramAggregators. It would probably take the full config instead of buckets as a separate argument. But overall, not much change I think.