New rule: Nullable val can be non-nullable

[BraisGabin]

Expected Behavior of the rule

Detect variables and properties that are declared as Nullable but they never contain a null value.

Non-compilant:

val a: String? = "hello"

val a: String?
   get() = "hello"

Context

The tanslator of code from java to kotlin adds a lot of unnecesary ?. And when you are refactoring maybe changes a property that could hold null to something that couldn't hold null but you can forget about redeclaring the type.

I don't know if we could handle var too.

After reading the last commits of #4353 I relasied that this rule shouldn't be that difficult to implement.

[severn-everett]

I think the case where var could be flagged is if the analyzer could prove that any values that it gets assigned are non-null (or at least not nullable). Private vars - along with vars that have private setters - seen pretty feasible for this; vars that are settable outside of their containing object/class might balloon the complexity (and/or computational time) quite a bit. In any case, I'll take a look.

[cortinico]

Closing this as it was shipped in #4379