[META] Run PHPStan on Drupal core

In the child issues there is already work done, and even some subissues split off which fixed things, spotted by phpstan, even it's not running on drupal core (yet).
I think in this main meta issue we should first discuss the plan of approach here, and link the correct issues. First step offcourse is to add phpstan, and then start fixing this. To break this up in smaller chunks, i would propose the following steps to doe.

1. Add phpstan config to core (level 0). Enable it on CI, and in pre-commit-script, do not fix any problems, just add a baseline. This way we don't introduce any new phpstan level 0 errors. But can keep this part focussed on adding and running phpstan.Code for this can already be found in subissue, but might need some rescoping.
2. Create a meta to fix all phpstan level 0 errors from the baseline. Some could probably be done in bulk, others might need seperate issue.
3. When all (or most) of level 0 items are fixed, update phpstan to level 1, and include a new baseline
4. Repeat step 2 and 3 till we're at level 8.

If this sounds a bit okay, i can work on cleaning up the subissues to follow this plan.

i think #5.1 makes sense especially now that PhpStan 1.0 is out (actually, 1.1 already).

I suggest to start tagging 'phpstan' all issues related to this, and 'PHPStan-X' with X=level all issues specific to solving a set of a PHPStan level problems.

I don't have the time to open a proper issue, but here's one we need to open. Replace usages of static::class . '::methodName' to [static::class, 'methodName'] array callable versus concatenation on static::class

That is why we have this error:

		-
			message: "#^\\#pre_render value 'non\\-empty\\-string' at key '0' is invalid\\.$#"
			count: 1
			path: lib/Drupal/Core/Render/Element/Pager.php

From:

  public function getInfo() {
    return [
      '#pre_render' => [
        static::class . '::preRenderPager',
      ],

It's just a PHPStan thing where static:class is handled a specify way for generics. So concatenating is weird. So let's use the proper array callable format for these scenarios and it will pass. See https://github.com/mglaman/phpstan-drupal/issues/301 for more background.

Other example: modules/big_pipe/tests/modules/big_pipe_regression_test/src/BigPipeRegressionTestController.php

After reading the baseline, I am tracking the following three issues to resolve a few false-positives or bugs in phpstan-drupal

* Function twig_escape_filter not found #310
* BrowserTestBaseDefaultThemeRule ignore InstallerExistingConfigTestBase tests #311
* RenderCallbackRule causes error in PlaceholderGenerator #303

Thanks @mallezie for cleaning up this meta, it helps a lot!

Decide when to go a level up? Does the baseline needs to be fully clean?

IMHO not necessarily, but for sure we need to reduce the ignoreErrors patterns to a minimum - either solving them or moving them to baseline. Without that we will be potentially silently adding new failing code without knowing - whereas with baseline we would be able to take a picture and avoid new failures. Also for sure the baseline right now is too fat.

Added issue for #14
#3259716: Replace usages of static::class . '::methodName' to first-class callable syntax static::method(...)
And added some stuff to the issue summary.

For moving up i fully agree currently our baseline is too fat. I think at least following requirements should be met:
* IgnoreRules and IgnoreErrrors are removed from the config (and fixed or in the baseline)
* The baseline should contain less than X items (first wild estimate for a number should be around 50, but depends on what is in the baseline). Could be a combination between less than X items in the baseline, and less then Y different messages in the baseline.

Just checked, Level 1 would add over 800 more errors ATM.

Linked #3293933: Upgrade phpstan/phpstan to 1.8.1

I suggest we close this at this stage, PHPStan is already running on level 1.