[cleanup] Remove three unreachable PHPStan type mappers

[TomasVotruba]

PHPStanStaticTypeMapper picks a mapper with a linear is_a() scan in registration order:

foreach ($this->typeMappers as $typeMapper) {
    if (! is_a($type, $typeMapper->getNodeClass(), true)) {
        continue;
    }

    return $typeMapper->mapToPHPStanPhpDocTypeNode($type);
}

So a mapper for a child type has to be registered before the mapper for its parent type. Three were registered after theirs, and can never be reached:

Mapper	Handles	Shadowed by	Position
GenericClassStringTypeMapper	GenericClassStringType	ClassStringTypeMapper	#13 vs #8
SelfObjectTypeMapper	SelfObjectType	ObjectTypeMapper	#30 vs #25
ThisTypeMapper	ThisType	StaticTypeMapper	#34 vs #31

None of the three is injected anywhere — they appeared only in the LazyContainerFactory list — so no call path reaches them and removing them changes no behavior.

They were also lying about what happens

Two of them describe an outcome that differs from what actually runs today. Feeding each type to the dead mapper and to the mapper that really wins:

GenericClassStringType, phpdoc:  dead mapper => class-string<Foo>    actual => class-string<\Foo>
ThisType, return position:       dead mapper => self                 actual => static

The live behavior looks like the correct one in both cases (fully qualified name in a docblock, static for a @return $this), so this is stale code rather than a bug — but it is the kind of stale code that reads as authoritative.

Guard

Added TypeMapperOrderTest, which walks the registered mappers and fails when one is shadowed by an earlier one. Reverting any single deletion makes it fail with:

The "Rector\PHPStanStaticTypeMapper\TypeMapper\ThisTypeMapper" is registered after
"Rector\PHPStanStaticTypeMapper\TypeMapper\StaticTypeMapper", but "PHPStan\Type\ThisType"
is a "PHPStan\Type\StaticType". It can never be reached, register it earlier.

That ordering constraint was previously enforced by nothing, which is how these three went unnoticed.