Usually the best alleged application is using them to do physics simulations that wouldn't be possible on a normal computer. Its a bit unclear precisely what that will open up, but normal computers opened up a lot of applications by being able to simulate classical system, perhaps QC will do the same for quantum physics things.

You're right that most of the other algorithms are pretty useless. Shor and Grover are certainly fascinating intellectually, but i doubt would be particularly useful (shor because everyone will switch to pqc if qc ever becomes relavent and grover is too slow to ever be useful)