refactor: Die on plugin communication error

[vcastellm]

All plugins uses safe code for error handling so in this case we can not do anything but die, exactly the same as in processor plugins.

Related to #652

[davidgengenbach]

@Victorcoder This will result in the main process getting killed when there was an error executing on the client (= the plugin) - as far as I understand?
This also means that a lot of time can pass before the plugin-death will be detected (= for jobs which are executed every week or so).
It may be interesting to add periodic health-checks to check for the plugin-health - and proceed to die when the plugins are not healthy.

Just an idea!

[vcastellm]

@davidgengenbach plugins processes should be considered as being the same thing as the main process, if one dies the main process should die fast as you said.

I will go with this PR better than nothing and I will left #655 open as an improvement.