Introduce budget for the Telemetry collectors

[mshustov]

Telemetry API allows plugins to query the Elasticsearch server to retrieve stored metadata. In some cases, an Elasticsearch response might have an enormous size as we saw in #115217 that affects the Kibana performance due to the necessity to parse JSON and perform some math operations over the data.

This issue is created to start an investigation on possible solutions to minimize the impact of telemetry on Kibana and Elasticsearch performance. As discussed on October's SDH issues grooming, we might need to enforce more strict rules for querying the Elasticsearch server from Telemetry collectors by introducing a budget for such queries. Telemetry collectors already have dedicated Elasticsearch and SO clients, so we have an extension point to adjust both client configurations to restrict access to API causing performance problems.

Before we take this task on, we need to check the metrics for telemetry collector queries on Cloud to see how far this problem affects our customers.

[elasticmachine]

Pinging @elastic/kibana-core (Team:Core)

[mshustov]

closed in favour of #119468