[CodeQuality] Add rule to move attributes under comment

[TomasVotruba]

The attributes are part of code, while docblock is comment above them. The other order will crash.

This was addressed in PHP FIG Per coding style:
php-fig/per-coding-style#8

And php-parser: nikic/PHP-Parser#762

Looking at code sample, maybe we could deal with this mess with Rector rule :)

-#[Attribute]
 /**
  * @property-read $name 
  */
+#[Attribute]
 class SomeClass
 {
 }

[samsonasik]

Stmts/Exprs that have attrGroups:

\PhpParser\Internal\PrintableNewAnonClassNode;
\PhpParser\Node\Param;
\PhpParser\Node\Expr\ArrowFunction;
\PhpParser\Node\Expr\Closure;
\PhpParser\Node\Expr\ClassConst;
\PhpParser\Node\Expr\ClassLike;
\PhpParser\Node\Expr\ClassMethod;
\PhpParser\Node\Expr\EnumCase;
\PhpParser\Node\Expr\Function_;
\PhpParser\Node\Expr\Property;

it seems PrintableNewAnonClassNode is for internal PhpParser.

[Crell]

Point of order: FIG has not addressed it yet. The question has been brought up to the working group. I agree putting the attribute after the comment is probably a good idea, but as of yet FIG has not made a statement one way or another.

[TomasVotruba]

Sure, it is motivation to move forward. We've seen this bug since first native attributes release.

[TomasVotruba]

After some reasearch and experiment, it's clear the php-parser resolves attributes correctly, even if separated by comment:
rectorphp/rector-src#2474 (comment)

So there is no need for such rule.