Works done: After fruitlessly poring over books on zeta functions, it seems Freeman Dyson's sotto voce nudge to classify generalized one-dimensional quasicrystals is a way to go. As he writes:
Question: Will this be a worthwhile strategy to pursue where a big picture akin to Atiyah-Singer index theorem needs to be made for symmetry?
The proof of the pudding is in the eating -- if following the program leads to a proof of RH, then ipso facto it is a worthwhile strategy. If it doesn't lead to the RH but does leads to other interesting things, it can reasonably be said to be worthwhile anyway. If it leads nowhere, then it wasn't worthwhile.
Which of those are the case nobody can say until it's actually been tried.
What kind of answer were you expecting? Anyone who can answer either of
should be out there winning an Abel prize for themselves, not doodling about on MSE.