As for me I encountered several times where i've said "ERRRKKK need to restart everything !!" because no pause on TestExecute.
Sometimes when you need to do lot of manual settings of environment before launching long lasting test and you miss a little thing or it has changed somewhere a little but that implies that you KNOW the test will fail then a pause execution would be so helpful.
And to manage the case of lphilippe_macau do you think the QA team know in advance when the the maintenance occurs ? I don't think so. Moreover on large system you have intelligent maintenance that can occurs at any time when the system detect the need of (self healing provisioning system).
About people connecting to machine and fix ... Yes of course, the target is to know why it's going wrong but the goal is to deliver test results within time so pausing will help to achieve the goal and AFTER they will investigate on erro to get rid of it.
Last thing is i think it's a big mistake assuming that TE is always under a full devops environment with tons of tools. TE is also used in fully functional environment (end user environment) without these tools. Lot of my customers have only a TE running on their system without any other tools (because of budget, because of system constraints, because of ..).
That was my 5 cents 😉