std::rand: Add an implementation of ISAAC64.

This is 2x faster on 64-bit computers at generating anything larger
than 32-bits.

It has been verified against the canonical C implementation from the
website of the creator of ISAAC64.

Also, move `Rng.next` to `Rng.next_u32` and add `Rng.next_u64` to
take full advantage of the wider word width; otherwise Isaac64 will
always be squeezed down into a u32 wasting half the entropy and
offering no advantage over the 32-bit variant.

src/test/bench/core-map.rs

@@ -109,7 +109,7 @@ fn main() {
         let mut rng = std::rand::IsaacRng::new_seeded([1, 1, 1, 1, 1, 1, 1]);
         let mut set = HashSet::new();
         while set.len() != n_keys {
-            let next = rng.next() as uint;
+            let next = rng.gen();
             if set.insert(next) {
                 rand.push(next);
             }

src/test/bench/core-set.rs

@@ -60,7 +60,7 @@ impl Results {
             let mut set = f();
             do timed(&mut self.random_ints) {
                 for _ in range(0, num_keys) {
-                    set.insert((rng.next() as uint) % rand_cap);
+                    set.insert(rng.gen::<uint>() % rand_cap);
                 }
             }
         }
@@ -102,7 +102,7 @@ impl Results {
             let mut set = f();
             do timed(&mut self.random_strings) {
                 for _ in range(0, num_keys) {
-                    let s = (rng.next() as uint).to_str();
+                    let s = rng.gen::<uint>().to_str();
                     set.insert(s);
                 }
             }

src/test/bench/shootout-fasta.rs

@@ -76,7 +76,7 @@ fn make_random_fasta(wr: @io::Writer,
     wr.write_line(~">" + id + " " + desc);
     let mut rng = rand::rng();
     let rng = @mut MyRandom {
-        last: rng.next()
+        last: rng.gen()
     };
     let mut op: ~str = ~"";
     for _ in range(0u, n as uint) {

src/test/run-pass/morestack6.rs

@@ -69,8 +69,8 @@ pub fn main() {
     let mut rng = rand::rng();
     for f in fns.iter() {
         let f = *f;
-        let sz = rng.next() % 256u32 + 256u32;
-        let frame_backoff = rng.next() % 10u32 + 1u32;
+        let sz = rng.gen::<u32>() % 256u32 + 256u32;
+        let frame_backoff = rng.gen::<u32>() % 10u32 + 1u32;
         task::try(|| runtest(f, frame_backoff) );
     }
 }