Add token rate limits to ensure that no one proxy token can consume more than its fair share of the overall GitHub App limits.

Every response from GitHub will carry data about rate limit usage. Rather than pass those values back to clients, we should:

• store the data for Prometheus metrics exporters to surface
• maintain an internal attribution of API usage per proxy token
• enforce our own limits on proxy tokens

The exact approach here is not defined. But there is a need, we now have a shared pool of requests that can be made over a time horizon and we need to ensure no one actor or group of actors can starve out another.