Don't recrawl identical URI/resource pairs.

Shaves ~10% off the suite runtime, even though we don't have more formal
benchmarks in place.

Author: Julian

referencing/_core.py

@@ -169,8 +169,9 @@ def with_identified_resources(
         uncrawled = self._uncrawled.evolver()
         contents = self._contents.evolver()
         for uri, resource in pairs:
+            if uri not in self._contents or self._contents[uri][0] != resource:
+                uncrawled.add(uri)
             contents[uri] = resource, m()
-            uncrawled.add(uri)
         return evolve(
             self,
             contents=contents.persistent(),