My psychic colleagues think this might be https://github.com/elastic/kibana/issues/58749 which is fixed in Kibana 7.6.1.
No I just attempted to retrigger a high CPU usage and as such the early stack trace isn't the same as latest search req sample.
execution_hint... properly above me knowlegde currently 
Quite possible that the missing time out on KQL auto completion could be the cause combined with User' impatience.
Will update to kibana 7.6.1 and cross fingers