It strikes me that Nik has most of the unique amp voices covered with his excellent clones - maybe it's time to branch out and do something like 65amps have done and create something entirely new using his knowledge of great tone?

For me it would be a great Fender clean channel and a Plexi crunch channel, probably the two most recorded sounds in all types of music. The challenge is to combine an amp which relies on a lot of clean headroom with one that doesn't. Rivera have probably come closest, so I'd go for a low-power 'boutique' version of that concept, but it could just as easy be a JCM800 with Dumble or JTM45 with AC30.

Cathode biasing would expand the flexibility, so you could use 6v6s or EL84s for recording and swap in some 6L6s or EL84s if you need more power for live work.

My dream Ceriatone would be 5E8A clean channel with Plexi crunch channel. Cathode biasing, tube rectifier, tube buffered FX loop. I've used both for recording, but when you try to A/B them for gigs, they're too different to balance up properly. Putting them both in one amp to share the same power circuit would even them out. That's why the 18/36 TMB with EF86 works so well.

I think you've already isolated the problem.  The problem is the Plexi doesn't get it's crunch for the preamp, power tube saturation (and speaker choice) is key to the signature "Marshall Tone." 

To pair a clean amp and (non master) dirty amp, you usually want the clean amp to be twice the output of the dirty.  For instance, if you use a 50-watt Plexi for dirty, then you want a ~100 watt amp for clean. 

Anyway, mixing to two would be fairly unusable.  Multi-channel amps usually rely mostly on preamp OD for thir dirty tones, it's the way 90-95% of multi-channel or high-gain amps are set up.

To get Fender and Marshall to work the way it needs to in one amp, you would need to have the dirty channel halve the output, either by switching to triode mode or switching off tubes.  The switching gets complicated fast. 

The problem with the kitchen sink amp you ask for is it does everything OK, but nothing great, you have to compromise something all-around to get decent tones from each channel.  Hence the Rivera, which is probably as close as anyone will ever get and have a profitable amp. 

That's why no one makes them.

Instead Dumble, Soldano, Bogner, Mesa, Egnator, etc. all had to do their own thing, all preamp based. 

Overwise, use small amps, and mic them both.

Yeah, I think you could be on the right lines with triode/pentode switching. Mesa have those options for each channel on the MKV I believe.

This is why I posted the idea - the easy route, as you say, is preamp distortion for any amp which can claim to have a decent clean channel. Surely there must be another way? Biasing? Cascading power tubes? Power Scaling?

I don't think there's as much of a gap as people think between pure clean and a decent crunch. Sure, if you want your clean-dist range from Mark Knopfler to Pantera, you need the MV the big amp manufacturers use to cover a lot of ground. But we're talking about something for session guys, so they only need take one amp to any session or gig.

I read a very interesting article written by Hartley Peavey. His claim was that the "power tube distortion" many seek is not just power tube distortion. It is also distortion created in the output transformer. So if you accept this, you would need to switch between e.g. two/four power tubes and at the same time switch between two different output transformers when going from clean to distorted, and vice versa.

He discussed an interesting effect occuring in the OT, called a dynamic bandpass filter. As the magnetic field approaches saturation, a bandpass effect occurs. This bandpass effect depends on the amplitude of the signal going in to the OT. He claimed that this effect together with the power tubes distorting at the same time, was the ultimate tube amp power sectio distortion.

The article ed with the Peavey's TransTube tech, but most of it is interesting stuff about tube amps:
http://www.peavey.com/support/technotes/hartley/chapter_3.pdf

It would be hard to get this dynamic bandpass effect in the OT, as well as getting a loud clean sound from the same amp, unless two different sized OT's are used...