Most mathematicians believe RH is true, and generally when doing industrial number theory people operate under the assumption that RH is indeed true and so if they need to use X to justify something and there is a theorem of the form "if RH is true then X" they use X.
Thus a proof of RH is not a problem. It just confirms that what people applying number theory already assumed was correct.
A disproof means that those X's might not be true and their use would need to be reassessed.
The congruence of squares equivalence to factorization demonstrated we need at least 500 bits and then the special number field seive that built on this push it to 1024. The general number field seive pushed it again to 2048.
Sure it’s not a log(n) break but it’s been broken. If you look at the complexity analysis of the special vs general number field seive the portion of the exponent going from 1/2 to 1/3 should give you thought. Can it be moved to 1/4? Could it be moved indefinitely to 1/x? The general number field seive is relatively recent. If someone comes up with a similar breakthrough again (and this has happened many times over with rsa) your 2048bit keys won’t be secure just as your 128bit rsa keys from the past are no longer secure.
A possible mitigation for things like websites would be either ECC or even using the quantum resistant encryption systems (the industry would more likely avoid this due to the systems being very prototypical since we have just started researching this).
Since old bitcoin wallets can’t be moved off of RSA you can transfer the coins to your wallet and there is no mitigation.