Magnetic reed switches can have maximum operating frequencies in hundreds of times per second, which on a bike wheel would be hundreds of miles per hour. So I think that having the magnet at the rim will make no difference whatsoever with respect to the magnet going by too fast for the sensor to read. If anything, I'd expect better performance at the rim because the faster attenuation of the magnetic field would mean less debouncing issues.

Reed switches have several advantages over hall effect switches, less power draw, cheaper, simpler, unaffected by electrostatic shock. The hall effect might last forever, but are more sensitive to current in wires, ESD and so on. Since a reed switch can last for billions of cycles, that's not really a problem. Convert a billion turns of the wheel into miles, and you'll see what I mean.

I think that the orientation of the magnet is a bigger concern than its position.