One small part to improve in the Cern 10 mA article would be to not use TaN resistor arrays, but NiCr based ones. There is a chance that this change was also done in thestep from version 2 to 3 at Cern. They actually had a comparison of resistor arrays done at Cern and found usually more excess noise for the TaN types.
The product / schematics may be proprietary. There is still nothing really secret, new. The kind of amplifying current mirror is also used with DMMs, like in the HP34401, though usually for lower currents and maybe for a different reason (to get stable current over a larger range) and not so much with the heating problem.