The heatsinking on the pill isn’t the best. You’ll probably need to fabricate a new thicker LED shelf as well that would wick the heat away. However, once you have enough mass hitting the inner side walls of the thing, it should be able to handle the heat.
We (including myself) have been talking about reflectors, but I wonder if maybe some TIRs would work? In particular, for 9 emitters, maybe 3 Carclo narrow triples, arranged triangularly?
EDIT:
So take 3 of these:
along with 3 of these:
and arrange the 3 in a triangle?
I’d like to see that. But, will it all fit in the host that Matteus will be using?
I got the big head Courui a week ago, very nicely built light with an excellent anodized surface. Feels very polished and professional in hand. The lens is large, much larger than an HD2010 but smaller than a TK61. I’ll double check measurements on it directly.
The reflector has a large flat base to it, might be ideal for what you’re planning. Not ideal for a mod involving a single XM class emitter. In stock form with a hot de-dome mine is doing 190Kcd. Very tight beam profile, surprisingly so.
It’s a nice light, cheap enough, big head, big reflector, multiple cells in parallel for runtime, go for it!
With room to spare 
diameter inside bezel = 65mm, 20mm stars.
Circles in circles here
3x20=60. Got 21 leds handy 
Cheers David
it’s not that wide at the shelf where the emitters sit. Tapered head, you can’t bore it that wide.
Now THAT would be AWESOME!
My guess: bad beam, no significant increase in throw (compared to a single LED). I'd go for individual reflectors/TIRs/aspherics.
It would be cool if you could have the center and outer emitters separately controlled. That way you could have true throw and floody throw. That is what my scratch build will have, but the emitters will be in a single row.
Dale, the head is over 65mm inside dia to a depth of 35mm, more than enough for a multi led reflector/tir
I know, pics or it is not so Edit, the pics are not showing for some so added the links.
Diameter, what else do I use to show its over 65mm.
www.members.optusnet.com.au/pommietoo/courui-head-dia.jpg
Depth.
www.members.optusnet.com.au/pommietoo/courui-head-depth.jpg
And a bit of off topic led porn just cos I can.
www.members.optusnet.com.au/pommietoo/courui-LED.jpg
Must remember to clean that bit of thermal paste off.
Cheers David
Might also point out that the emitter shelf screws in, the head is bored through. So it’d be pretty easy to change the height of the emitter shelf, material, depth, whatever is needed for the case at hand.
68.2 inside the bezel, the shelf for the emitters that threads into the “pill” area where the driver mounts is 41.3mm. And yes, the head is 68.2mm down to around 35mm deep, then tapers to meet the top of the “pill”.
David, your pictures aren’t showing up over here.
In the first image, theres a Samsung 18650 laying nicely on the shelf to demonstrate it is over 65 mm wide down on the shelf. Although I can see its not the emitter shelf, simply the bit the pill/shelf screws into, it is the head itself.
Edit: 
Right, but Pommie is pointing out that it would be relatively easy to make it work. A new copper shelf, wider on top with a reduction then the threads would be ideal and this would allow height adjustment as well. The reflector could be shaved to fit or optics could be utilized to further raise the shelf surface and give more copper.
I wasnt commenting on whether it will work or not. With enough persistence, almost anything is possible. I was just relaying the image and the fact you were measuring the pill area (shinny silver bit), he was measuring the head (annodised bit).
Here's a wall shot of my TR-1200 with five XP-G2. For some reason it gives me a huge hot spot.
And is that your “Ouchy” foot down there at the bottom, or the “other” foot?
^
ROFL
I agree. I have played with three leds soldered together, tight together and even five, but the problem is that as the beam gets out farther, black spots appear, in between the leds. De-domed would actually make that worse. The larger the total die surface, the larger the spot will be. No matter what reflector you use, you can never make an XM-L spot as small as an XR-E spot and you can’t make it as intense. The beam is going to enlarge the further away you get and as it does, it will turn more and more into separate beams, with “holes” in between.
Small die with high surface brightness. We need an XR-E on Steroids. I wish there was enough interest from Cree to make a new one, with their newer technology.
Might be easier to make a light with water cooling or something, so that an existing small die could be pushed to very high amps. Maybe even some kind of nitrogen cooling or something.
I tried two XP-G2 in a C8 reflector. The beam profile was not pretty.
Ouchyfoot has put his finger on it (see what I did there? ;-)). Several folks predicted that kind of behavior. A 3x3 array might have a slightly more pleasing beam profile due to having something in the center. Then again the XP-E2 uses a smaller die so die-to-die spacing would be worse than what Ouchyfoot showed. [I see OL was pointing out something similar]
Mattaus, you discounted the multi-aspheric thing earlier, but you might be surprised at how effective it can be. Packhorse and Trout/Troutie-mtb on CPF both had builds with multiple ahorton aspherics. 7 Up Ahorton Aspheric Lamp | Candle Power Flashlight Forum
That said, I don’t know how one of those measures up against a single XM-L2 setup.
You could only fit 3x of those particular aspherics into the Courui anyway.