Consider adding unrelated_type_checks like with unrelated_type_equality_checks

[ltOgt]

The following is covered by unrelated_type_equality_checks:

void someFunc(int number, String string) {
  if (number == string) {} // linter gives a warning for unrelated types
}

But the following does not give a warning:

void someFunc(int number, String string) {
  if (number is String) {} // no warning for unrelated type check
}

[srawlins]

My 2 cents: I think there enough occasions where you want to check if a is a type that is unrelated to it's static type, so I would not report the general case.

But you can still report cases that will never be true, in the cases of final classes. The example above, "is this int a String?" is a perfect example.

Given the expression x has static type T, and you're checking x is U (or x as U, I believe):

• If T is final and U is neither a supertype of T nor a supertype of any of T's subtypes (which can only be declared in T's library), then the check will always be false.
• If U is final and T is neither a supertype of U nor a supertype of any of U's subtypes (which can only be declared in U's library), then the check will always be false. (I don't think I have this backwards...)

The same does not hold for sealed classes, as their subclasses can be subclassed arbitrarily.

We can similarly report unnecessary checks, where U is a supertype of T.

[FMorschel]

Inspired by this issue I created #5062 about unrelated_type_casts.