Yes, its really all about where the PI thinks it's getting key strokes from. I don't know about straight into VI or Pico, for two reasons:
1. There will need to be some kind of SIO response from the PI for it to participate as a SIO device. Of course, you could go straight serial through an R: driver, but then you'd need the linux serial TTY set up, which isn't hard to do.
2. Vi or Pico will be expecting a certain key mapping and set of commands. Somebody in between would need to map the atari keys and the atari program output ( such as print a character at position x,y ) to commands that would effect the same result on the screen.
As much as I recall<not much ) terminal screens pretty much had to exchange/refresh data for the entire screen each time when something BIG was done if there wasn't a good knowledge of the capability of the terminal/console. Some of the other programs like WordStar operated similarly in the CPM/DOS environments. That is, if you did something like <ctrl>Y to delete a line, the whole screen would clear and be redrawn with what the editor thought was current.
All the editor programs <IIRC> had built in help screens with useful information like ^Y = delete line.
In the Linux environment you can specify what terminal you have so it could go to the lowest common denominator like AD3, VT100, ???. You don't know know or care about what the Atari thinks because all of that matters is what the Linux box thinks and on a Pi, you would have that plugged into your composite monitor.
I think Kermit on the Atari was specifically designed to mostly work with UNIX, never spent more then a time or two using it.
I'm not saying it isn't nice. Shell account with access to all those neat utilities like telnet, Pico, Pine, C compiler, internet gateway, ogles of fast storage, giga hz of processing power. Problem for me is I start thinking 'Gee, I'd like to use that nice USB keyboard' and next thing you know, you are using the Pi natively.