Resolve staggerred container restart issue on waku-simulator host

[stubbsta]

The watchtower config is supposed to pull the latest nwaku image when it becomes available and then restart all containers which use that image.
However it seems that only a few containers are restarted, causing nodes of multiple versions of nwaku to be running.
The restart also causes the other nodes to lose peer connectivity and the first messages from the newly restarted node to be identified as invalid messages.

One suggestion is to implement linked docker containers to force watchtower to restart all applicable nodes:
https://containrrr.dev/watchtower/linked-containers/

[alrevuelta]

Unsure how the watchtower works, but can that run a script? If we can run a script every time a new image becomes available we can control the logic. eg stop all containers, clean whatever is needed, etc.