1. As other commenters point, many cases have intact memory and circuits for what would be considered "lost", which activate from time to time. So it's likely a question of SNR and tissue vitality (e.g. basic capillary function) for these cases.
2. You wouldn't believe what feats of neuroplasticity lie behind a few receptors properly pushed by molecular keys. We just don't have experience to describe it. Adult neuroplasticity and (disproven btw) neurogenesis is a rigid sad joke compared to what's possible.
OK maybe, although not having experience to describe it sounds somewhat worrisome for a scientific project.
where your point 1 is concerned, again, since the brain has many redundancies one can totally have brain tissue destroyed and with some slight therapy get back to normal functionality, but when Alzheimer's is too advanced I suppose you will not be able to fully revert, but maybe revert to some earlier stage - sure.
2. You wouldn't believe what feats of neuroplasticity lie behind a few receptors properly pushed by molecular keys. We just don't have experience to describe it. Adult neuroplasticity and (disproven btw) neurogenesis is a rigid sad joke compared to what's possible.