Improve executor startup checks

At the moment we have a Sgt. Hartman who performs a 100 boot tests. This is too much for executor restarts, fine for 1st seen checks though.

However, we need to do something, because often after maintenance tasks at the remote farms, things are not correctly plugged and so on. We should do a better job of alerting operators when this happens.

So, the proposal is to create a quicker sanity check, to ensure that machines boot twice for example, that we can collect their serial loggings, etc. Things to make sure the operators have connected all the hardware up correctly, as best as we can.