My Docan cells look like they will be fine with the JBD's roughly 200ma passive balancing now that I have increased the setting for the cell voltage at which balancing turns on from the factory default of 3.3v to 3.4v. My problem was that it was balancing the wrong cells when that voltage setting was too low.I think silicon is cheaper than nichrome or whatever these days.
But you will have to try harder not to burn it up. Good choice of resistor and power dissipation is limited. Transistor turned on too hard would melt. I've used PTC resistor to adjust bias of a transistor, but that thing (and 0805) cost me a buck.
But this is a fun intellectual problem, and could be useful for cells that are not so well behaved. If these can be built at $5 per cell in a bus bar form factor that you just put on top of each cell with no additional wires or additional bluetooth links that need to be watched, I like that better than adding a more complicated commercial device that does 2A active balancing. Yes, it's still just passive balancing, but is simple and solves the primary problem of having a few cells run up beyond 3.6v while others are still getting charged. Dump currents could be increased as high as might be needed for some especially problematic grade C cells by adding extra resistors in parallel.
Your earlier suggestion of using a PNP to drive the high power NPN is good, since your PNP's Vce-sat is lower than my NPN driver's Vbe. A little bit more headroom to get the driver's collector current dialed in.
Yes, I would have to try harder for appropriate proportional control of a transistor, switching is easier. Every commercial balancing circuit I have looked at does it with switching. Wirewound through hole resistors of appropriate size don't necessarily need a heatsink, here's a datasheet showing one that works fine at up to 70C ambient air temps before needing to be derated https://www.mouser.com/datasheet/2/303/res_tuw_tum-1228619.pdf Also, that resistor is built to deal with surface temps of 275C, where heat dissipation is much faster than it is for the 120C or so max case temp of a transistor.