Write a "how to add more stats" guideline document

[alvestrand]

This has been on the plan for a while, but it's nice to have a bug on it.

[henbos]

This document will be targeted towards developers and aims to:

• Give developers an idea how standardization works and guide them through it.
• Put ownership of the entire process of adding the stats (spec issue, spec PR, implementation) on the developer proposing the stats. Allowing resources and expertise to be used appropriately.
• Decrease the perceived and actual delay time of standardization efforts. Spec should not be a blocker unless we really should not add the stats.

No more "I filed an issue and nothing happened so I added a goog-stat".

This will allow us to:

• BAN adding more goog-stats, which will...
• Incentivize more development efforts to be put into spec-compliance, and
• Encourage switching to the spec-compliant API.
• And eventually deprecate legacy getStats().

[henbos]

The document should also cover experimental stats, both the kind we aim to standardize but need to test the metric before we know it's a good one and the kind that is only useful for Chromium debugging and evaluating the result of feature experiments.

[henbos]

Experimental stats will have limited exposure, e.g. not exposed to the web platform by default (hidden behind Origin Trials experiment, etc.).

[henbos]

Here's the doc I intend to share, please comment.
https://docs.google.com/document/d/1q1CJVUqJ6YW9NNRc0tENkLNny8AHrKZfqjy3SL89zjc/edit?usp=sharing
I'll send it to the mailing list tomorrow. @vr000m this also shows an overview of some of the C++ implementation of getStats() even if it doesn't tell you how difficult it is to add a particular metric.

[alvestrand]

Sent to mailing list, no comments. Guess it's perfect.

[alvestrand]

It makes sense to either link to it or copy it into the webrtc-stats repo - not as a spec, but as "helpful material".

[henbos]

I'll create a PR to link to it from our README.md.

[henbos]

We're so closed to PR we probably don't need this anymore

[vr000m]

Go to provisional stats to give input