Fix integer vector mul/div operators and bindings.

* Vector2i and Vector3i mul/div by a float results in Vector2 and Vector3 respectively.
* Create specializations to allow proper bindings.

This fixes #44408 and supersedes #44441 and keeps the same rule of int <op> float returnig float, like with scalars.

core/math/vector3.h

@@ -343,6 +343,9 @@ Vector3 &Vector3::operator*=(const real_t p_scalar) {
 	return *this;
 }
 
+// Multiplication operators required to workaround issues with LLVM using implicit conversion
+// to Vector2i instead for integers where it should not.
+
 _FORCE_INLINE_ Vector3 operator*(const float p_scalar, const Vector3 &p_vec) {
 	return p_vec * p_scalar;
 }