test performance please

performance test scheduled: 1 job(s) in queue, 0 running.

This is slightly different from scala/scala#6582 which does not change how eligible candidates are computed but rather does a second pruning check right before calling typedImplicit. It'd be good if we could come up with more testcases to evaluate the performance of each approach. See discussion on Gitter: https://gitter.im/typelevel/general?at=5ae775d162316e0505176c8c

(I've commented out most of inductive-implicits-bench.scala to get it to pass the CI with a normal stack size and -Yno-deep-subtypes enabled, but we should add the full thing to our benchmark suite)

Performance test finished successfully:

Visit http://dotty-bench.epfl.ch/4426/ to see the changes.

Benchmarks is based on merging with master (75b66c8)

Oops, I just realized this patch doesn't have the effect I thought it did on inductive-implicits-bench.scala, closing until I figure things out properly.

Superceded by #4447