Fix test for multi_qubit_measurements/state_modification (#1830)

Doing Reset on the first qubit before checking the state of the second
qubit can lead to empty solution being judged as correct (with low
probability). We have to check the state of the second qubit first and
only then reset the first one.

katas/content/multi_qubit_measurements/state_modification/Verification.qs

@@ -37,20 +37,21 @@ namespace Kata.Verification {
                 // Prepare the state to be input to the testImplementation
                 StateInitialize_StateSelection(alpha, qs);
 
-                // operate testImplementation
                 Kata.StateSelection(qs, ind);
-                // reset the first qubit, since its state does not matter
-                Reset(qs[0]);
 
-                // apply adjoint reference operation and check that the result is correct
+                // Apply adjoint of state preparation operation
                 Adjoint StatePrepare_StateSelection(alpha, ind, qs[1]);
 
-                if not CheckAllZero(qs) {
+                // We only care about the state of the second qubit; 
+                // if it's still entangled with the first one or not in zero state, this check will fail.
+                if not CheckZero(qs[1]) {
                     ResetAll(qs);
                     Message("Incorrect.");
                     Message($"The state of the second qubit for {params}, ind = {ind} does not match expectation.");
                     return false;
                 }
+
+                Reset(qs[0]);
             }
         }
         Message("Correct!");