feat(perf): Simplify poseidon2 cache zero-pad (#5869)

# Description

## Problem\*

Resolves <!-- Link to GitHub Issue -->

Works towards general Brillig bytecode size reductions

## Summary\*

When actually setting up the state for a poseidon2 permutation, we have
a loop to zero-pad the cache and then another loop to add the cache to
the state. If we just have a single loop that only adds to the state
when the rate index is below the cache size we achieve the same goal
with less code.

## Additional Context



## Documentation\*

Check one:
- [X] No documentation needed.
- [ ] Documentation included in this PR.
- [ ] **[For Experimental Features]** Documentation to be submitted in a
separate PR.

# PR Checklist\*

- [X] I have tested the changes locally.
- [X] I have formatted the changes with [Prettier](https://prettier.io/)
and/or `cargo fmt` on default settings.

noir_stdlib/src/hash/poseidon2.nr

@@ -27,15 +27,13 @@ impl Poseidon2 {
     }
 
     fn perform_duplex(&mut self) {
-        // zero-pad the cache
-        for i in 0..RATE {
-            if i >= self.cache_size {
-                self.cache[i] = 0;
-            }
-        }
         // add the cache into sponge state
         for i in 0..RATE {
-            self.state[i] += self.cache[i];
+            // We effectively zero-pad the cache by only adding to the state
+            // cache that is less than the specified `cache_size`
+            if i < self.cache_size {
+                self.state[i] += self.cache[i];
+            }
         }
         self.state = crate::hash::poseidon2_permutation(self.state, 4);
     }