scrape (histograms): Implement OpenMetrics histogram experiment

[beorn7]

OpenObservability/OpenMetrics#247 describes an experiment to embed a preliminary text format into the existing OM text format to gather experience with native histograms as text format.

We need to implement parsing this experimental OM format (behind the native histogram feature flag). Note that this is not meant to be the final format. It is only an easy way to play with a text representation, which will then inform the final text representation.

[beorn7]

Also note that this is different from the effort to extend the OM protobuf format for native histograms.

[beorn7]

Since @fstab has implemented Native Histogram support in client_java by adding protobuf, the immediate need to hack some kind of support into the OM text format is somewhat diminished.

Furthermore, nobody else approached me about the experiment.

And finally, adding Native Histogram support to the OM protobuf spec seemed relatively straight forward, which gives implementers yet another option.

All of this shouldn't imply that we should drop the experiment, but I think we can demote it to P2 as there are enough options to play with Native Histograms beyond Go.

[beorn7]

Given the reaction to a brainstorming doc, I think we should not pursuit the "embedded JSON" idea any longer (but we can, of course, change our minds again). Of all the ideas discussed, "embedded JSON" (idea 1 in the doc) was the least liked, notably also by @csmarchbanks, who maintains client_python, which will probably be the first instrumentation library to implement a text format for native histograms.

Therefore, I'm closing this issue (for now).