So I think there’s plenty of potential for making a more informed decision now, rather than just assuming what we’ve already got is sensible.
Since the BIPs propose to leave the existing sigversions unchanged and instead aim to add a new sigversion (through a new tapscript leaf), which adheres to the varops budget and potentially implements GSR, it will not have any impact on those existing sigversions including the current base tapscript leaf.
Therefore, it is reasonable to construct a computational budget with parameters, such that the worst case script with GSR activated, is no slower than the current worst case for most or all machines.
From the DoS viewpoint, an attacker can choose any sigversion, so constraining the computational budget in a new sigversion further than the worst case in other sigversions, would not have any benefits, it would constrain the new sigversion unnecessarily.
Maybe you are arguing that the varops budget should also be added to the existing sigversions? That would be a completely different discussion imo.