Remove rate::ilog64()

[rzumer]

It is superseded by util.rs::ILog::ilog().

[lu-zero]

It exists because we cannot have const-fn traits in stable and that could be the case for a while =/

[rzumer]

Right, so one of #1465 and #1466 can be chosen, or alternatively we could add const fn functions for all required PrimInts besides i64, not via a trait but in the same way that ilog64() works now.

[coveralls]

Coverage decreased (-0.004%) to 81.303% when pulling 467a545 on rzumer:ilog into cdb51aa on xiph:master.

[rzumer]

After discussing with @lu-zero he feels indifferent about it (correct me if I'm wrong) and I prefer this solution, since less duplicated code seems a bigger benefit than constifying a non-critical function. I will squash and push after CI succeeds.

[lu-zero]

It is better to have less lines of code around until the const-fn lands completely.

Keep in mind that it has an effect if the function arguments are constant themselves, otherwise it doesn't do anything :)