fix: Long startup time when OnEvent 'start' times out (repeatedly)

[Nerivec]

@Koenkk I don't believe there is a scenario where onEvent should take precedence over the startup of the rest of Z2M? So this should be fine. Should actually help with spammy networks.

Fix #25681
This particular issue (probably more with the same cause though) was tracked down to this change:
Koenkk/zigbee-herdsman-converters#8479
These devices are apparently rather picky, and can be offline for a while after the coordinator comes back online, which means, if you have a lot of these, and they all time out, you end up with (10s timeout * number of devices) right in the middle of startup.
Same applies for other devices with similar scenario (hence the change at the top logic instead of in every converter).

@antst thanks for the very detailed testing for this, and the (long) time taken! 😉

[antst]

I can add, that with one of my devices, which lost route, while coordinator was not really online for quite some time due to testing, and stopped to respond completely, it even didn't time our after 10 sec, it sent message and was waiting for reply forever (I tested it up to 20 min), blocking start of z2m for this time.

[antst]

Just in case, don't merge it yet.
I will perform some definitive tests in few hours, to see, it check if this change is ultimate solution.