K2: Avoid using Nothing? as inference result in the majority of cases

Namely, do not choose `Nothing?` result type when fixing a variable
that has other constraints besides the ones that came from
the relevant type parameter's upper bounds.

See more details in KT-55691.

In K1, the case from specialCallWithMaterializeAndExpectedType.kt
was working (inferred to String?) just because the branches
were analyzed independently with `String?` expected type.

This change became necessary after the previous commit when we united
inference subsystems for if/when branches (see motivation there).

NB: For K1, the behavior is left the same, but the code
was refactored a bit.

^KT-55691 Fixed
^KT-56448 Fixed

compiler/testData/diagnostics/tests/inference/nothingVsParameterBound.kt

@@ -0,0 +1,15 @@
+// FIR_IDENTICAL
+// SKIP_TXT
+
+class Out<out T : CharSequence?>(val t: T)
+
+fun foo() {
+    // We have two constraints here:
+    // Nothing? <: T (from argument `null` type)
+    // T <: CharSequence?
+    // And we fix T to `Nothing?`, because it's still more preferrable than constraint from the upper bound
+    val x1 = Out(null)
+    bar(<!DEBUG_INFO_EXPRESSION_TYPE("Out<kotlin.Nothing?>")!>x1<!>)
+}
+
+fun bar(w: Out<String?>) {}