Replace the quadradic duplicate check logic with a linear version

It's ~2% slower with 10 hashes and ~2% faster with 100 hashes (with very
noisy benchmarks). But the important part is that the runtime should
scale linearly with the number of inputs. E.g., this version is 2.5x
faster for 1000 signatures.

fixes #68

src/signature.rs

@@ -1,4 +1,4 @@
-use std::io;
+use std::{collections::HashSet, io};
 
 #[cfg(feature = "multicore")]
 use rayon::prelude::*;
@@ -139,12 +139,15 @@ pub fn verify(signature: &Signature, hashes: &[G2Projective], public_keys: &[Pub
     // Enforce that messages are distinct as a countermeasure against BLS's rogue-key attack.
     // See Section 3.1. of the IRTF's BLS signatures spec:
     // https://tools.ietf.org/html/draft-irtf-cfrg-bls-signature-02#section-3.1
-    for i in 0..(n_hashes - 1) {
-        for j in (i + 1)..n_hashes {
-            if hashes[i] == hashes[j] {
-                return false;
-            }
-        }
+    if hashes
+        .iter()
+        // This is the best way to get something we can actually hash.
+        .map(|h| G2Affine::from(h).to_compressed())
+        .collect::<HashSet<_>>()
+        .len()
+        != hashes.len()
+    {
+        return false;
     }
 
     #[cfg(feature = "multicore")]