The more I think about it, the more I'm sure a fork will be needed. The exploits must be patched. Better to do it with as short of a timechain as possible
Discussion
You don't even need a fork; in fact, a fork is unfeasible for both DDoS attacks and spam, but I have to admit that attacks are becoming increasingly subtle.
All of this will end with John the carpenter asking you why he can't run a node with a $1,000 computer, and then you'll know it's all over.
I have to admit, I have no idea what the requirements of a fork would be or what the consequences are. Mythinking is that if Segwit & Taproot have existential defects, they WILL eventually be exploited.
If the reference implementation (Core & associated mining software) allow for the exploit then it's only a matter of time.