I "think" if you grind the concave side on a surface grinder (which it is understood that you do not have nor particularly want for your space & circumstances, but it is the way I can think about it) that the bow will either increase or stay about the same. I don't "think" it will bow the other way.
To increase the odds of a deeper bow in the existing direction: Put the file on a couple/three thicknesses of typing paper to protect the chuck. Concave side up. Use full magnetic power to warp the file down as flat as possible.
Use the Z axis table traverse, no coolant and a dull wheel. run the table in & out at the center and then work back & forth to each end of the file, dropping the wheel and useing the full width, toward each end of the file, letting the work get just as hot as you like. The ends will most likely pop up off the chuck and you can continue until the bow suits, or the file flys off the chuck
.
I'm enjoying the incongruous concept of trying to grind (sort of) flat work so it bows, so no doubt this is coming across a bit flip. But it is not intended to be rude. It is IME as good a recipe as any for your objective.
An analagous situation would be to use an old large diameter sandstone wheel, hold the ends of the file with something so you can stand the heat, and grind the full width of the back with no coolant, press hard.
OTOH, for either method, there is going to be a fine line between optimum heat for the best shrinkage, and softening the teeth on the file.
I'll be curious to hear of any experiments.
I would take (have done) the approach someone else mentioned, go to fleamarkets, stores, cabin fever, wherever NOS files are sold and pick the ones most warped if that is the preferred condition for a certain task.
Thanks for the compliment!
smt