reconsider fuzzing and randomness as possible mitigations

[npdoty]

We had previously concluded that randomization was unlikely to be helpful and actually recommended against it. But some browsers have some new experience with deploying random values with some success. Should we reconsider this point?

https://w3c.github.io/fingerprinting-guidance/#a_standardized_profile