Metric names

[NicolasLM]

It seems that currently metrics do not fully follow Prometheus best practices. It makes it hard to know what the metric actually refers to (ie. allocation_cpu_throttle is in fact a duration).

A few documents explain in great length how to name metrics:

• https://prometheus.io/docs/instrumenting/writing_exporters/#metrics
• https://prometheus.io/docs/practices/naming/
• https://promcon.io/2016-berlin/talks/so-you-want-to-write-an-exporter/

According to that I propose the following names:

• nomad_allocation_memory_rss_bytes
• nomad_allocation_memory_limit_bytes (and switch the unit from MBytes to Bytes)
• nomad_allocation_cpu_percent (prom seems to favor ratio instead of percent, but this keeps things consistent with nomad terminology)
• nomad_allocation_cpu_throttled_duration_total_nanoseconds (may be switched from a gauge to a counter since it seems that this value cannot decrease)

Let me know what you think about this proposal and if you want me to submit a PR.

[Nomon]

sorry for the late response, nothing against confirming to the prometheus best practices. Pull requests most welcome.