Anyone remember Rainey's ROC chassis that was so stiff, they had to go to something else in the middle of the season? or Edwards removing the upper motor mount bolt in his VTR in his last year of WSB?
Keep in mind, as some of these guys have said, it's about desired flex in the desired direction. I believe you'd want torsional ridgidity so the frame doesn't twist (think about an old chevelle ss that's been highly modded and the twist will be obvious) but you would also want some lateral flexibity without allowing so much flex that the frame is like a wet noodle. A good example of that concept is a racing swingarm - tall and thin so it doesn't twist or flex front to back, but has a slight bit of float side to side for when the bike is leaned over. a problem with allowing these things to flex is how do you control their motion without dampening? some of it can be done with shape, but some of these things don't liked to be flexed over and over or you can get work hardened parts, right? not an easy dilemma to solve and often, stiffening one part of the bike, causes another to flex or softening one spot might do the same. i believe if you have the correct input force and just the right amount of flex, you can induce natural frequency oscillations and i'm guessing that that is the chatter the hondas and yamahas have had.
to answer your question, a car doesn't lean over as much when it turns so your desire is to maintain as much down-force on all four tires by limiting the over-turing moment created by the CG - Lower the CG, reduce the amount it moves rotationally