So my proposed mitigation strategy would be:

1) General measure against any 50+% attack: Leave the door open for a change in the mining algorithm. Don't concentrate on a single algorithm like Scrypt, making it more difficult for any attacker to prepare for this event. Perhaps even maintain a fork of Bitcoin code with several other algorithms, so the switch can happen rapidly.

2) Specific measure against a "rival blockchain attack": Identify a third blockchain directly competing with the attack blockchain (in the case of Ethereum being the attack chain, for example Solana, Cardano or Avalanche). So those invested both in the attacked and the attacker blockchain can dump their stakes on the attacker blockchain buying the third chain's coins, reducing the attacker blockchain's value and increasing the third blockchain's value. The third blockchain's whales will very likely not participate in the attack, because they will benefit much more if the attack blockchain crashes and they can get the market share.

3) Extreme last resort: I would not be against just preparing the ground for a fork with PoS (e.g. with a proof of concept or even a testnet "what could happen" if PoW really fails), which would be never enacted. But the possibility alone should disincentive any attack based on supposed specific PoW vulnerabilities. If #1 is a standard nuclear bomb (can be employed in rare cases like Hiroshima/Nagasaki), this would be the H bomb (will never be used but it's advantageous to have it).