Some of this is off topic but may be of interest.
As for running the adapters for EL84, it may fit in a combo but not a head. No space. Mark V will not work with either (assumed) you could inquire about this through email to Mesa as they should be able to support the product as it is still in production.
In my opinion, the Mark V90 sort of stepped off the Mark amp norms, good or bad. It just does not have the characteristic I would expect for a Mark amp.
You lucked out, you bought a Mark V90. If you wanted to run 6V6, the Mark IV in tweed power (same as variac) was possible.
Since they upgraded the power for the extended class A circuits to 45W, that places the center pair in the hot zone. I assume that is in Pentode configuration. Ch3 can switch to triode so it is no longer running at 45W, more like 22W.
I had compared the Mark V90 to the Mark VII as it is quite different amp in terms of its power section. Not only does the Mark VII sound amazing (it better for that price) it has some similar traits to the Mark V90. At full power it pushes a good peak at 140W rms. Just like the Mark V90 and also similar to the JP2C (100W class A/B). Not a slouch in power by any means. Totally different preamp design that is well balanced on all modes as well as a good match to the JP2C in the IIC+ and IV modes. The Mark V90 just does not get there, too many preamp tubes and attenuators to compensate for the extra gain stages.
Anyways, at 90W both Mark VII and V90 are about the same power level wise. Most of the post gain distortion is due to the center pair in extended Class A operation. Mark amps will utilize power tube saturation and distortion, not exactly a Dumble but close enough. At 45W, again both are at the same power level. On CH3 of the Mark V, change from pentode to triode and you end up with a drop in power level down to 22W to 25W, well that is what the Mark VII does, no more 10W mode but has a 25W power mode, runs the center pair extended class A in triode.
You almost get a bit more with the Mark V90 since you have tube tracking for CH1 and CH2, pentode/triode mode for 90W or 45W (this only affects the center pair). Then there is the 10W power mode.
To be honest, I would rather use the Mark VII over the Mark V90. It just sound better and I Can do more with it. However, the V90 may need some slight mods but still working out what should be changed. I did have to adjust the bias just so I could run Mesa STR440 tubes, before the change it would red plate them in no time. SED =C=6L6GC or Tung Sol 7581 were the only tubes that could be used without instant death.
The Mark IV in general could run a mix quad of 6L6 and EL34 like the Mark III simul-class. It too had a pentode/triode switch for the class A tubes. Also, there was no bias switch for tubes, so running a full set of EL34 was not possible.
You can view the Mark IV manual on Mesa's website if this interests you.
EL34 power mode, this has some good effect but will run the screens much higher in voltage. Gold Lion KT77 did well at full power. However, the manual states when using EL34 power tubes, you should keep the amp in Variac power for longer tube life. Since there is no voltage regulation or suitable voltage drop (typically the screens will have Zener diodes to act as voltage drops to reduce the screen voltage relative to the plate). So, the reason for variac power use is to reduce the overall screen voltage and keep it at a safe operating level. Most amps that are designed for use of EL34 like the Badlander or Triple Crown will have zener diodes in the screen circuit. Once you convert over to 6L6, it sounds like it is lacking power. It is all due to the screen circuit.
Also, on the circuit of the Mark V90, there are two zener diodes that get pulled into the screen circuits when using 10W power mode. After looking at the schematic for the Mark V90, there is a lot of things going on for each power mode. Not easy to keep track of what relay does what.