Affix driver rim to aluminum, and remote temp sensing?

Long-time lurker (12 years?), first-time poster.

I’ve been considering making a custom P60 drop-in or three…or five…and play around with trying to make a 2CR123/1xLIon buck driver (if nothing else, I’ve got a neglected SDS1100X-E that needs some love).

For 20mm, I kind of like the design of the aluminum CWF P60 pill. The 4-40 screw hole I can work with, but what about the rim of the driver PCB? With copper or brass, it’s easy enough, with some cursing and patience, so solder a few points, leaving a secure PCB, that can still be removed. But, is there any similar option for such an aluminum pill, outside of potting? On the off-chance my driver development endeavor is eventually successful, I’d want to take them apart and rebuild them.

Unrelated to that is LED temp. The ideal place to check temps for the LED temp is on the same PCB, right nearby. The Mountain Electronics 20mm dual-channel MPCBs can be made to work for that, but none others that I’ve seen. The few other Optic-based P60 pills available (AFAIK, only Mountain’s) are big copper slugs, that will heat up very slowly. Are there any readily available ways to get readings right near the LEDs, using other DTP MCPCBs?

As an alternative to a temp probe, is Vf temp change relatively constant, in most cases, or does it vary, like Vf itself? Approximations of datasheet curves would probably be easy enough to implement, with some one-time self-calibration (IE, run the driver at moderate current to a steady state at temp X for a couple minutes, and then record those Vfs, across a range of currents, to the uC’s flash, with many points along the curve baked into the firmware).