I spoke to some manufacturers and they seem open to the idea of unibody designs. The only concern is that it is a little more expensive to machine since regular flashlights can be machined from short stock for the head and pipes for the bodies, as opposed to a single long stock with a lot of material cut away.
Today I managed to open the ZL SC700d, and reassembled it back after (it's challenging for those without the right tools so I don't recommend anyone doing so without a risk of damaging it) and it's nicely designed; they used a PIC microcontroller, a TPS61088 regulator and PCB-side-mounted e-switch, all on a single but fairly large PCB (basically one inch square).The main inductor is a Coilcraft 7030 1uH, a perfect choice for the 61088. The design is very similar to most of the switching driver circuits including the Lume drivers, so I guess I'm glad that this shows the Lume is on the right path. I can see ZL swapping out the 61088 for the 61288 for future flashlights for even more power, but I think the Lume X1 is still the first driver to use the 61288 !
The LED is mounted in a channel cut in the PCB where a copper block provides thermal contact and heatsinking to the body. The regulator and inductor are potted, but in a soft silicone. It's compact, but perhaps too compact - and here comes my main criticism. The design does come at the cost of difficulty of reparability and warranty for the product, though it's robust enough that I think the failure rate is going to be quite low.
This also makes LED swaps difficult since the design is intrinsically package and footprint specific. It doesn't really save much room either since it uses a separate PCB as the positive-battery-spring contact (using pogopins), so it still uses 3 boards in total, just like regular flashlights if you count the MCPCB (driver, positive spring, tail spring), but it does improve thermal performance.
Regardless, I think there is still some value in a 3-piece design (head, body, cap) - it's cheaper to machine in general, and I guess it's required for easy re-flashing without using a ridiculously long pogopin programmer (going with the AVR-1 and 3-pin UPDI makes this almost possible, though). I'm not sure how many manufacturers will want to go with the casting method - perhaps there are other possible options but I'm not familiar with them. The unibody does improve on thermal sinking, though.
Unibody is one thing, the quality of the machining and anodization is another. I think so far Emisar/Noctigon comes the closest. I think the main thing that strikes me about the ZL is the economy of design - it's so compact and tidily built and I see no reason why other flashlights can't be designed similarly even if the machining isn't as good. I'd like to keep the LED and MCPCB swapping though, since that's really part of the ethos of the BLF community, and I think all these are not difficult to achieve at the same cost of machining.