Set up cava benchmark

• Set up cava benchmark w/ criterion.
• GPU schedule for cava.
• Convert a bunch of benchmarks from using hercules refs to use hercules boxes.
• Define AlignedAlloc global allocator in hercules_rt that aligns everything to 32 bytes.
• Bump nom version.

hercules_samples/dot/src/main.rs

@@ -4,14 +4,16 @@
 use hercules_rt::CUDABox;
 use hercules_rt::{runner, HerculesImmBox, HerculesImmBoxTo};
 
+use aligned_vec::ABox;
+
 juno_build::juno!("dot");
 
 fn main() {
     async_std::task::block_on(async {
-        let a: [f32; 8] = [0.0, 1.0, 0.0, 2.0, 0.0, 3.0, 0.0, 4.0];
-        let b: [f32; 8] = [0.0, 5.0, 0.0, 6.0, 0.0, 7.0, 0.0, 8.0];
-        let a = HerculesImmBox::from(&a as &[f32]);
-        let b = HerculesImmBox::from(&b as &[f32]);
+        let a: ABox<[f32; 8]> = ABox::new(32, [0.0, 1.0, 0.0, 2.0, 0.0, 3.0, 0.0, 4.0]);
+        let b: ABox<[f32; 8]> = ABox::new(32, [0.0, 5.0, 0.0, 6.0, 0.0, 7.0, 0.0, 8.0]);
+        let a = HerculesImmBox::from(a.as_ref() as &[f32]);
+        let b = HerculesImmBox::from(b.as_ref() as &[f32]);
         let mut r = runner!(dot);
         let c = r.run(8, a.to(), b.to()).await;
         println!("{}", c);