Or terminate it at Dartford and restart from there depending on the lateness, there's a number of places on the SouthEastern where you can terminate short of the final destination and still have alterative services for passengers to use.
I can only re-iterate what was posted earlier, which is that control will be looking at a slightly bigger picture.
You can't just terminate a train short if the train is booked to have a crew relief at the terminus. I'm not familiar with the crewing arrangements on the Rainham service, but I have a hunch there's a crew depot at Gillingham, which would rule out turning short on at least some services.
Controllers will seek to look at the situation they're faced with, and come up with an immediate plan to "sort" the problem. They ideally want to be able to make a decision there and then, cascade it out to those people who need to know, and then forget about it. That's the only way to keep afloat if they're, for example, faced with a whole list of uncovered driver duties, trains which need to go to a particular depot, stock balance gone awry, or whatever. If things subsequently change then they will simply treat this as a new problem to solve.
In your late running example, the controller will look at what the train *and* the driver do, and decide upon a plan. There's essentially four options:
(1) If there's scope for the late running to correct itself naturally, then do nothing.
(2) If the train doesn't have a crew relief at the terminus, then turning short is an option.
(3) If the train does have a crew relief at the terminus, missing stops may be the best option. If so, there's a secondary decision as to which stops to omit. Controllers won't necessarily have intimate local knowledge of what the dominant local passenger flows are, so rightly or wrongly it will probably be an operational decision based upon getting the service back, rather than considering the immediate needs of the people on that service. Naturally the ideal is to do as little as possible to get things back, or at least give the chance of the train being on time for its next trip taking into account the turnround time *and* what the driver is booked to do next.
If you don't want to do this option on the immediate trip (for example because it's a heavily loaded key peak service, or because there's already been issues and it will create too much of a gap for certain stations) then the controller will look at what the driver does. If he's on the train for a while then let it run late, and sort out on the next trip by any one of the recovery methods. A judgement will need to be made on whether it's worth letting the late running continue for the sake of avoiding inconveniencing the current batch of passengers.
(4) If there's a batch of crew reliefs at a given location, it might be possible to reform the trains into on-time ones. Normally this will involve making up one or more trains on time, and producing one very late train which will then be heavily turned short or miss its circuit altogether. This is most effective when stopping patterns, stock type and destination are the same. The more variation there is, the less likely this method can be used. One thing to beware of is where the trains are booked to finish, for example one train might be on its diagram specifically because the depot want it back - the closer things are to end of service the more this is an issue as there's less time to put things right again.
This method can be quite passenger friendly as if done right some passengers might not even be aware it's happened -- although we did have an example on this forum when such a set of reforms was very neatly done but someone complained that it messed up their seat reservation!
Why not just terminate at Watton at Stone and run it ECS to Hertford CS? Especially as there should be staff there to terminate and lock up the ECS services while they run up to Langley Jcn to reverse.
I suspect the answer to this is more simple, the controller was seeking to run as much as possible with the resources available. If you're cancelling something then rule number one is don't cancel any more than you need to. Without sounding flippant, the object of the exercise is to run the timetable ... it’s a cop out to cancel part of a service for which you have a train and driver present as booked.
But what do they actually consider? If it's the fact that the train needs to make up time then just lose the Harpenden stop as Parkway and St Albans are the most important stops being very busy apart from Luton on the Kents north of the river as Parkway is there for the airport and St Albans offers connections for intermediate stations plus a walk across the town to use the Abbey line.
As alluded to elsewhere, don't assume the controller will have an intimate local knowledge. The main focus is to sort the problem, in this case that is to get the train and driver back on time as soon as possible, without imposing a negative effect on other services. In an ideal world it may well be worth putting a couple of minutes on a couple of other services in order to get one very late train back, but PPM regulating doesn't really encourage this. So in reality he'll be looking for an option which isn't going to mess up anything else. With the plan being seemingly to route the train on the fast line, a stop at St Albans would risk delaying other services far more than gliding through at speed.
Another point to bear in mind is the controller might well not have time to sit down and devise a perfectly crafted plan, in reality he's probably got a whole list of problems to "sort". We on here have the luxury of spending time thinking about what might be the best possible solution for one particular problem in a perfect world, with the benefit of hindsight. The controller's role doesn't allow that luxury.