Helpers for migrating from `telemetry_prometheus(_core)`

[hauleth]

I have just noticed, that PromEx library do not properly set bucket sizes for Peep. They are using "old" buckets option.

So the question is how we should work with that. Should Peep library dynamically construct modules for these or maybe it should warn about that. Maybe both?

[rkallos]

I'm not very confident in my answer because I feel out of touch with indirect users of Peep.

Warning seems a bit impolite, because I don't think users of PromEx can do much about the issue directly.

I think it would not be too difficult to make Peep 'just work' for PromEx. We could add (back) support for :buckets, and Peep can maintain an association between distinct values of :buckets option values and their matching generated Peep.Buckets.Custom module. Since Peep expects its metrics at startup, creating these modules can be done as part of traversing the passed-in metrics.

Another thought is what to do when a VM runs multiple instances of Peep; do they share the same registry of custom buckets modules, or do we accept the possible waste of multiple generated modules for the same :buckets value? As far as I am aware, running multiple Peep processes in a larger release is not something that happens frequently, so the simpler approach of one set of custom bucket modules per Peep process seems reasonable.