Infrastructure-as-code for running lance.blue

docs: what a match costs, and what the hung ones cost master

Runbook section on collecting and reading per-match cost, and the two TODO entries it can now answer with numbers: 14 of 69 matches ran over three hours and account for 92% of the $49 spent, and Container Insights is 0.7% of that while being the only surviving source of a task's peak memory.


+48 -3
3 changed files