Rename process.runtime.jvm.* namespace to simply jvm.*

[trask]

From @jsuereth #20 (review):

We want namespacing to be "reasonably unique" and minimal.

This will help with keeping the attribute names under control as well, now that we are namespacing them also.

[trask]

comment from @tigrannajaryan #20 (comment)

What is our general strategy with regards to choosing top-level namespaces? Is every language free to put anything in the top-level namespace (like Java does with 'jvm.`)? My concern is that this is going to result in highly crowded top-level that is going to be hard to manage. How do we make sure languages don't step on each other's toes? A language can choose to use a namespace and later it turns out to be also needed for another language but can mean something different.

I would like to understand what the policy is in general and I don't see it mentioned anywhere.

[trask]

I would like to understand what the policy is in general

my guess(?) is that it is up to the discretion of @open-telemetry/specs-semconv-maintainers to approve any new top-level namespaces.

@tigrannajaryan are you looking for something more than this?

[jsuereth]

@tigrannajaryan may still be out on Holiday. His request, in the TC, was that we set a consistent definition of what makes a namespace and what doens't.

E.g. we should haven't a metric namespace runtime.process.jvm and an attribute namespace jvm. We should consistently have namespace rules so that like things have like namespaces.

THis is where the "reasonably unique" language came from, we just need to codify it in a PR.

[liustanley]

Are there any future plans of making the runtime metric prefix consistent across the different language instrumentations? All of the other language instrumentation libraries use process.runtime.[language].

[trask]

hi @liustanley, I think this is a good question, can you open a new issue for it?