Optimize matcher creation in stores

[fpetkovski]

Is your proposal related to a problem?

Creating regex matchers in stores is currently quite expensive, at least according to a Receiver CPU profile. We have a lot of alerting and recording rules, so query throughput is fairly high.

Describe the solution you'd like

We could pool matchers or use some sort of an LRU cache to avoid the massive churn caused by high query RPS.

Describe alternatives you've considered

N/A

Additional context

[bwplotka]

👍🏽 Help wanted!