Hard state change CRITICAL to WARNING notifies immediately instead of following max_check_attempts


Just noticed that the max_check_attempts is followed only for OK to WARNING, WARNING to CRITICAL state changes.
While recovering the recovery notification is sent immediately when the state changes from CRITICAL to WARNING. How can I make this recovery also to behave like max_check_attempts option.

Icinga 2 version:
icinga2 - The Icinga 2 network monitoring daemon (version: r2.13.2-1)
Icinga2 Master and Node server OS: Ubuntu 20.04

Are you sure about the check attempts being considered from WARNING to CRITICAL? My understanding is, that only from service_problem 0 to 1 check attempts are considered.