[NI] Make subtyping algorithm more robust to error types

During subtyping/incorporation we transform types (e.g. changing nullability,
 form of the type) and, basically, we're doing this to some FIXPOINT.
 It's important that we use `KotlinType.hashCode()` to compare types, but
 for error types hashCode is a hashCode of its supertype and, for example,
 `makeNullableAsSpecified` method recreate type every time. So, we continue
 to generate new constraints and we'll never stop incorporation algorithm

compiler/testData/diagnostics/tests/inference/constraints/errorUpperBoundConstraint.kt

@@ -0,0 +1,26 @@
+// !LANGUAGE: +NewInference
+
+// FILE: Sam.java
+
+public interface Sam<K> {
+    Sam.Result<K> compute();
+
+    public static class Result<V> {
+        public static <V> Sam.Result<V> create(V value) {}
+    }
+}
+
+// FILE: Foo.java
+
+public class Foo {
+    public static <T> void foo(Sam<T> var1) {
+    }
+}
+
+// FILE: test.kt
+
+fun test(e: <!UNRESOLVED_REFERENCE!>ErrorType<!>) {
+    Foo.foo {
+        Sam.Result.create(<!DEBUG_INFO_ELEMENT_WITH_ERROR_TYPE!>e<!>)
+    }
+}