Numeric `BitFlipMutator` only flips 1/8th of bits

[stuxnot]

Hi!

Thank you for all your work on LibAFL!

While looking at the numeric mutators, I noticed that the BitFlipMutator uses size_of::<I>() as the upper bound for the flipped bits index, but afaik size_of returns the byte size of the type.

LibAFL/crates/libafl/src/mutators/numeric.rs

Lines 272 to 273 in df17dc2

	let offset = state.rand_mut().choose(0..size_of::<I>()).unwrap();
	input.flip_bit_at(offset);

This would make the mutator only ever flip 1/8th of the possible bits (assuming all bits are used).

Since I was implementing Numeric for a custom type, I'd love to see a fix that extends Numeric with a bits() method instead of multiplying the result of size_of by eight.

I'd be more than happy to create a PR for whatever solution you prefer.