Witness selection should pick exact matches for effect overloads

You can overload a function based on its `async`-ness, and
resolution is carried out based on async-ness of calling
context.

But during protocol conformance checking, for an `async`
requirement we were accidentally choosing the non-`async
overload instead of the `async` one. The `async` one is
the choice people would expect, since the `async` requirement
is in essence the "context" that forwards to the underlying
witness. This intended behavior is also inferred from:

#40088

The problem boiled down to a bad check when categorizing the
witness matches prior to ranking them.

Resolves rdar://109135488 / #60318

Author: kavon

lib/Sema/TypeCheckProtocol.cpp

@@ -1196,9 +1196,13 @@ swift::matchWitness(WitnessChecker::RequirementEnvironmentCache &reqEnvCache,
       matchKind = MatchKind::RenamedMatch;
     else if (requiresNonSendable)
       matchKind = MatchKind::RequiresNonSendable;
-    else if (getEffects(witness).containsOnly(getEffects(req)))
+    else if (getEffects(req) - getEffects(witness))
+      // when the difference is non-empty, the witness has fewer effects.
       matchKind = MatchKind::FewerEffects;
 
+    assert(getEffects(req).contains(getEffects(witness))
+               && "witness has more effects than requirement?");
+
     // Success. Form the match result.
     RequirementMatch result(witness,
                             matchKind,