AI Experimentation Framework

I have been tweaking the experimentation framework for years now, but recently I made a branch that gives it a complete facelift.

(This is just a test run)

I love being able to do more than A vs B, and I love the graph and the data table, and I love being able to see what the experiment is doing at a glance. I also like that I made a rule for Firebender to launch experiments with it, and that it can pass in a synopsis of what it is trying to test.

I also strongly defined different experiment types, and what flags should be passed in:

• Match CI - 100% cached
• Match CI - Typical feature dev cache bust
• Match CI - 100% cache bust (P90 - convention plugins ABI change)

The custom flags I made to make every build the same as a fresh CI build are

• --clear-gradle-caches
• --delete-file-hashes
• --delete-project-cache
• --delete-transforms
• --clear-os-cache (Linux caches files that you read so that it can read them faster next time!)