Reduce allocations in ProtoClient codec

[codefromthecrypt]

I was trying to understand the proto client and noticed some things that could be improved, if you are pretty familiar with how okio works. I ran the ProtoExample afterwards and it has the same behaviors.

Note: The upstream ApiClient isn't managing resources very well, particularly ensuring responses close. So, even though I adjusted in proto, there are other paths which can leak responses, especially on exception.

[codefromthecrypt]

@brendanburns I don't want to swamp you with review work, but if you are keen to have a "benchmarks" module which uses JMH, one day I'll add it. That way, you can see before/after and we don't need to guess if allocations are better or worse than before. It works similar (enough) to go's bench, even if not as convenient as isn't built-in.

[brendandburns]

/lgtm
/approve

The proto client hasn't been worked on in a long time and the protocol buffer objects haven't been regenerated for an even longer time. So if you want to take on improvements in this area, contributions are very welcome, but there's definitely going to be rough edges in the current experience.

Here's where the protos are generated:
https://github.com/kubernetes-client/gen/tree/master/proto

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: brendandburns, codefromthecrypt

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [brendandburns]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment