Yeah, that would be tricky to merge existing ones together. But you should be able to fine-tune the inpainting model the same way as you tune with dreambooth?
For faces, I haven't looked deep, but seems CodeFormer's training cost is minimal, should be able to fine-tune that model, probably better?