Using a Nernst-cell would be the logical solution. Infinite lifetime, high accuracy, immune to water or other chemicals...
Unfortunately only runs at 500°C!
When we are talking about rebreather compatible we are also talking about low power consumption. Heating the whole sensor does not strike me as a solution.
Blowing diluent over the sensor... that should work (I think the Mk5 works this way), but I dont know what flow you will need. Anyway, needs carefull design and is not easy at all (I dont want 2l of dil blown into my loop every 5 secs). Measure O2 optically? I dont know which wavelength you will need. If the vibrational modes of the O2 molecule are in the UV you have the problem of a suitable light source.
Besides, I said it is tricky, not impossible
Greets
Dirk