Postgresql Instance Metrics stop hitting PMM-Server and is not consistent

During testing for 2.27.0 we found that PMM-Server ends up hitting scrape timeout for `custom.hr` metrics for PostgreSQL instances, resulting in Empty dashboards like attached images. 

STR:

1) Add remote PostgreSQL, and keep standard scrape configuration.

2) Check postgresql instance summary dashboard, let the instance run for some hours, you should not see any gaps in metrics

3) if you check /prometheus/targets end points, on the unhealthy section you could see that some of the scrapes are timing out, thats mostly related to custom.hr

User Impact: Empty Postgresql Dashboards.