Seems like it's everywhere these days. Even in urban metro lines where you've got like 1km between stations. HK subway trains have been doing this since the 90s. Wikipedia sez Delhi Metro has it too.
(And even if you don't give the energy back to the network and just burn it in the resistors in locomotive, it gives you more effective braking. Locomotives from 1960's already had that.)
You could save some energy by using only recuperating electrodynamic brake to zero to stop the train and store it in some supercapacitators (not on the train, of course) if there isn't any other train to use it. Because you have many powered axles, you could brake quite effectively and fast enough. Well, possibly still not fast enough to use just EDB.
Raising top speed is of course going to cause much higher aerodynamic drag, so if the energy use is primary concern, it's not exactly the right thing to do. I rather meant eliminating any drops in the average traveling speed except the stations.
But I can't really argue against energy savings, because most likely, there would be energy savings :-) If you could built such a system safe enough (I don't believe that) and with less than absolutely insane costs (I don't believe that either :-)).
Yes, but accelerating the entire train is going to cost way more energy than accelerating the transfer car. Also, the transfer car can simply use wheels and brakes to gain kinetic energy from the train, so it can be incredibly light. Even if a longer track and motor are necessary for high speed bullet trains, there would still be a big energy savings.
Also, no matter how quickly you can accelerate to make a stop quickly, the same train will be able to save considerable trip time with the transfer car system. Saving a half hour or hour off of trip time from a route is certainly worthwhile, especially if it means that the train can run the same route at a lower top speed. The energy savings from that could be enormous over the long term.