Enterprise Metrics and Observability

Customers have expressed a desire to get detailed metrics about the performance and usage of Tilt. There are two concrete use cases for this information:

1. I'm an infrastructure engineer working on developer tooling. I want to know if a change that I made to a part of our developer tooling process designed to improve performance had the intended impact. I want to see graphs of the performance of Tilt, broken down in a fine grained way: ie, how long does it take to apply k8s YAML, how long does it take to build a specific image, etc.

The first use case also overlaps with a use case we have: we want to be able to know if the changes we are making are having their intended impact, and what are the most painful parts of the user experience on a per customer basis.

2. I'm an infrastructure engineer and a user has written in with a problem. I want to see, for that user, their history of using Tilt as it relates to performance and deploy history so that I can verify the problem their reporting and learn more without having to take up their time unnecessarily.

Again this is also a problem we have: see the amount of back and forth necessary to diagnose performance problems in GitHub issues or in #tilt.

No files, tests, or entry points are named. Start by mapping how Tilt currently records build, Kubernetes YAML application, performance, and deploy activity. Done means there is a defined approach for fine-grained performance graphs and per-user performance and deploy history.