I am sure other will chime in here soon about what is available (I know there are some specific controllers, just not sure of what they are called right now).

But I have wondered if a PLC would be up to the task as well? Or would it be 'too slow'?

Most custom pin makers are using the P-ROC boards/system.

Its a really nice system that controls everything. It flexible enough to handle EMs to color VGA display.
http://www.pinballcontrollers.com/

I wrote a program that uses multiple libraries to handle different needs.

-wrote a directsound library to handle sound effects and looping sounds, fading, panning, etc
-used an older version of fmod to handle mp3 streaming playback / jukebox mp3 playback
-wrote libaries to handle parallel port output, solenoid protection routines, lamp/flasher output
-wrote a bunch of directdraw code to handle the graphics (2d only) and some basic effects/animation abilities etc.
-wrote the gamecode core structure
-added a graphical console module that can be brought down with ~ to debug things (looks like a console from a game like quake 2 or 3 (etc))
-switch handling library, programmed to ignore repeated switch hits in rapid succession / vibrating switch.

This project was started well before the P-Roc and all those things were popularized, so I've been sticking with it so far. It's based on the pinmame-hw design. Who knows, I may change things up in the future. It's just for fun. Writing all the libraries and making sure they behaved nicely was time consuming, but now it's nice to have pre-built libraries I can play with. Initially it started as a very complex visual pinball table (I should have released it, it went well above what VP originals were doing at the time) and then a complete re-write into a standalone EXE with COM+ objects happened from there.

The way it is now, I have a machine with pretty serious sound capabilities, nearly endless mode capability, good graphics potential with an LCD monitor in the head (fairly easy to develop for but somewhat time consuming), PC controlling the whole thing; it's quite a capable platform to develop on if you have some older pc's you can sacrifice.

Main problem now is waiting for some replacement CNC parts to arrive so I can start getting to work re-designing and cutting the playfield (old one was a mockup cut by hand)!

I would second the P-ROC if you don't want to build your own hardware.

Another option is to do what they did, and create your own processor board that connects to existing driver boards.

One step further, which I think is what you want, would be a full redesign of the hardware/software. Basically, you can use schematics for existing platforms to see what is necessary and tweak the design for your needs. Keep in mind that the mainstream platforms were designed intentionally generic so they could be used for any game.

Could be anything. I like C, Assembly, and Verilog/VHDL. Your processor will read switch inputs and output to transistors for coil/lamp driving. The actual processor can be pretty much anything with enough speed and I/O. Since you are looking at auxiliary ICs for sound and video, a microcontroller would fit your needs. In my opinion, the ideal solution would be an FPGA with cores for sound/video/output driving, and then a microcontroller core to run the show.

My best advice is to keep the hardware/layout rather simplistic. I've never seen a strong correlation between popular games and the amount of "stuff" on the playfield. Players seem to be mostly about the rules (software).

Let me know if you want any help, and keep us posted!

Yup, use the P-Roc for your game. I have a p-roc setup and a custom C++ engine for game logic/sound/video/etc that I use for many of my custom builds (I'm not a fan of Python script).