Rename libafl_bolts::rands::Rand::zero_upto to below_or_zero. (#2911)

Hi LibAFL!

I was playing with the [`Rand`] trait when I realized that the
documentation of [`Rand::zero_upto`] did not match what I was expected:

fd6271fa35/libafl_bolts/src/rands/mod.rs (L139-L142)

When using the following RNGs, [`Rand::zero_upto`] never returns the upper bound `n` as it would have been expected according to the documentation:

 - `RomuDuoJrRand`
 - `RomuTrioRand`
 - `Sfc64Rand`
 - `XkcdRand`
 - `XorShift64Rand`
 - `Xoshiro256PlusPlusRand`

The default implementation of [`Rand::zero_upto`] is to use [`fast_bound_usize`],
which excludes the given upper bound, thus I believe here that the default implementation
of [`Rand::zero_upto`] is wrong.

As discussed here: https://github.com/AFLplusplus/LibAFL/pull/2911#issuecomment-2623773829,
we believe that renaming the method would be better than changing the actual
implementation.

[`Rand`]: fd6271fa35/libafl_bolts/src/rands/mod.rs (L108)
[`Rand::zero_upto`]: fd6271fa35/libafl_bolts/src/rands/mod.rs (L139-L142)
[`fast_bound_usize`]: fd6271fa35/libafl_bolts/src/rands/mod.rs (L100-L103)

libafl/src/mutators/mopt_mutator.rs

@@ -608,7 +608,7 @@ where
 {
     /// Compute the number of iterations used to apply stacked mutations
     fn iterations(&self, state: &mut S, _: &I) -> u64 {
-        1 << (1 + state.rand_mut().zero_upto(self.max_stack_pow))
+        1 << (1 + state.rand_mut().below_or_zero(self.max_stack_pow))
     }
 
     /// Get the next mutation to apply

libafl/src/mutators/mutations.rs

@@ -73,7 +73,7 @@ pub fn buffer_set<T: Clone>(data: &mut [T], from: usize, len: usize, val: T) {
 pub fn rand_range<S: HasRand>(state: &mut S, upper: usize, max_len: NonZeroUsize) -> Range<usize> {
     let len = 1 + state.rand_mut().below(max_len);
     // sample from [1..upper + len]
-    let mut offset2 = 1 + state.rand_mut().zero_upto(upper + len - 1);
+    let mut offset2 = 1 + state.rand_mut().below_or_zero(upper + len - 1);
     let offset1 = offset2.saturating_sub(len);
     if offset2 > upper {
         offset2 = upper;

libafl/src/mutators/scheduled.rs

@@ -145,7 +145,7 @@ where
 {
     /// Compute the number of iterations used to apply stacked mutations
     fn iterations(&self, state: &mut S, _: &I) -> u64 {
-        1 << (1 + state.rand_mut().zero_upto(self.max_stack_pow))
+        1 << (1 + state.rand_mut().below_or_zero(self.max_stack_pow))
     }
 
     /// Get the next mutation to apply

libafl/src/mutators/tuneable.rs

@@ -132,7 +132,7 @@ where
                 iters
             } else {
                 // fall back to random
-                1 << (1 + state.rand_mut().zero_upto(self.max_stack_pow))
+                1 << (1 + state.rand_mut().below_or_zero(self.max_stack_pow))
             }
         } else {
             // We will sample using the mutation probabilities.

libafl_bolts/src/rands/mod.rs

@@ -136,8 +136,8 @@ pub trait Rand: Debug + Serialize + DeserializeOwned {
         fast_bound(self.next(), upper_bound_excl)
     }
 
-    /// Gets a value between [0, n]
-    fn zero_upto(&mut self, n: usize) -> usize {
+    /// Gets a value below the given one or zero
+    fn below_or_zero(&mut self, n: usize) -> usize {
         fast_bound_usize(self.next(), n)
     }