no. i mean samples.
you start with one state: b0 and b1 not asserted.
you capture the inputs at some interval (say, 1000hz).
after 10 samples, if the new state got more samples than the previous state, you change states.
congratulations, your input is debounced.
this requires exactly 1 timer, will work up to a huge number of inputs, and will not prefer any input over another.
adjust number of samples, sample interval or whatever value will flip the previous input according to your requirements.