My lathe's magnet sits in a slight recess, which I guess gives it more grip on a machined flat surface and stops it sliding on a smooth round shaft.
Sliding may be a problem – although I can't imagine a strong magnet coming off, it might move sideways. Sideways movement may not show up in practice. Provided the magnet's close enough right-left to activate the hall sensor, rpm will still be measured. Even if a magnet gradually walks around the circumference, the effect on rpm readings will be small.
A spot of glue would also stop sideways movement and be less trouble to apply at home Another guess; it's cheaper for the manufacturer to machine a flat recess on a lathe spindle than it is to clean it carefully and apply a glue resistant to oil and heat. Also easier to fix an unglued magnet if untrained labour puts it on upside down.
Dave