query: Optimizations for 100+ stores

[bwplotka]

No one had use case like this, also it easy to mitigate with intermediate Thanos queriers, but we might get into issue while asking so many stores.

Acceptance criteria:

• Benchmark before and after
• We can control / limit memory consumption for case of 100+ stores.

Some idea: manage pool of go routines for gRPC stream instead of having each for 100+ stores with constant (1) channel buffer each)

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.