All the time working with Zabbix I wonder whether it could be possible to put some smart logic into the self-monitor/timer processes to prevent falsely generated events based on .nodata() trigger functions, when something goes terribly wrong - what is likely to happen due to the centralist approach

This could probably take internal events into consideration and/or be based on smart thresholds/profiles that may identify abnormal operation of Zabbix processes.

Now, the only way to achieve something similar to that (but by far not the same) is to create a central trigger that is based on internal items from which other nodata() triggers depend on.
I believe it's not necessary to explain the difference in usability, reliability and efficiency