Add a check that detects degenerate pool sizes
Broken out of #36377 (closed)
MR: !24265 (closed)
We recently ran into a subtle regression where the database pool sizing logic (that scales it up to the maximum thread count) did not trigger due to an initializer ordering problem: #199672 (closed)
We should detect these sort of regressions after all initializers have run and fail fast.
Edited by Craig Gomes