Take into account duplicate FQN from theories

src/Sample/UnitTest1.cs

@@ -2,6 +2,14 @@ namespace Sample;
 
 public class UnitTest1
 {
+    [Theory]
+    [InlineData(1)]
+    [InlineData(2)]
+    public void Test1(int value)
+    {
+        Assert.True(value > 0);
+    }
+
     [Fact]
     public void FailsOnce()
     {

src/dotnet-retest/RetestCommand.cs

@@ -170,7 +170,13 @@ Dictionary<string, bool> GetTestResults(string path)
                     var isFailed = result.Attribute("outcome")?.Value == "Failed";
                     var method = doc.CssSelectElement($"UnitTest[id={id}] TestMethod");
                     Debug.Assert(method != null);
-                    outcomes.Add($"{method.Attribute("className")?.Value}.{method.Attribute("name")?.Value}", isFailed);
+                    // NOTE: we may have duplicate test FQN due to theories, which we'd run again in this case.
+                    // Eventually, we might want to figure out how to filter theories in a cross-framework compatible 
+                    // way, but for now, filtering by FQN should be enough, even if not 100% optimal.
+                    var fqn = $"{method.Attribute("className")?.Value}.{method.Attribute("name")?.Value}";
+                    if (!outcomes.TryGetValue(fqn, out var wasFailed) || !wasFailed)
+                        // Only change the outcome if it was not already failed
+                        outcomes[fqn] = isFailed;
                 }
             }
         }