[alerting] public facing doc on alerting performance

[pmuellr]

note: this was split off of issue #95194, which was originally supposed to cover both benchmarking 7.14 and creating a blog post on alerting performance

We'd like to have some public-facing doc on alerting performance, similar to other elastic-generated content like this one covering performance with logs and metrics: Benchmarking and sizing your Elasticsearch cluster for logs and metrics. Doesn't necessarily have to be a blog post, but these seem to be popular destinations for stack users. Perhaps a split between some additional asciidoc and a blog post.

The basic idea is to describe:

• how the alerting system uses task manager at a very high level,
• describe how many tasks can be run per minute/hour given the existing task manager config defaults
• how to change the task manager defaults
• how to get measurements from your system (task manager health)
• how to determine if you need to resize your cluster based on more or less alerting rules running
• throw in some extra stuff like ephemeral workers
• should we publicize our stress tester, as something customers could use for experiments?

[elasticmachine]

Pinging @elastic/kibana-alerting-services (Team:Alerting Services)