Earlier quoted context omitted.
It has been considered and basically it’s not really reliable or practical or safe.
Is there any writings on this? It sounds interesting.
---
Reliability
For a connection like this to work the trains have to meet, consistently, every hour or however frequently they run. This is pretty difficult, most rail networks are highly complex and delays cascade across a network. With station transfers this is okay, you can drop off anyone who needs to transfer to wait and let everyone else continue, but in this sort of moving transfer either people just totally miss their connection because it wasn't there, or the train loops around somehow and delays everyone else continuing on.
---
Practicality
Let's say trains meet at 80 mph. Assuming you give people one minute to get their bags, walk to the door, take their seats in the other train (fairly aggressive for anyone who isn't a fit young adult) that means you spend 1.3 miles of distance traveling. That means 1.3 miles of parallel track where neither train can actually stop to serve local communities.
Station transfers in comparison are fairly compact (the length of a train) and they can actually also let people on and off from the surrounding areas with appropriate exits, whereas that's not possible with moving trains.
---
Safety
We can't really guarantee two trains will move at the same speed parallel to each other. Trains are not automated to such a degree outside of self-contained metro networks, full automation is too complex to do in one shot and partial automation is still very complex and in its early days. Also whatever physical mechanism has to be tight enough to accommodate accessibility (wheelchairs can't go over large gaps) and reliable enough to work all the time; what happens if said mechanism breaks down while the trains are moving and conjoined?
---
Finally there's the matter of actual need. If you need trains to not get slowed down, you can have some express services skip connecting to local regional services at all. It turns out demand for that kind of service is fairly limited; as an example, Amtrak has tried several times to introduce nonstop DC to NYC service, but it turns out that the additional passengers do not offset the loss of passengers from Baltimore, Wilmington and Philadelphia.
