Followup to #109345
Requires and includes #113580

Fixes #113588

Explanation:

• the original idea of this fix is to only set the path for non fully loaded scripts without caching them in the ResourceCache
• In the original PR I applied another change on top which tried to ensure, that fully loaded scripts would be cached by the ResourceLoader
  • In hindsight this does not seem to be necessary. And for cyclic references it did not work. Because the GDScript compiler cached the fully loaded script before returning control to the resource loader.
  • The resource loader only sets the path on resources if the resource is not already part of the cache. However the GDScript resource format loader did reset the path before returning control to ensure that Scripts get cached when the ResourceLoader sets it. In the cyclic reference case the ResourceFormatLoader did reset the path but the ResourceLoader did not set it again, because the script was already added to the cache by the compiler.
  • With this PR the ResourceFormatLoaderGDScript is once again responsible for adding fully loaded scripts to the cache (as it was the case before Prevent shallow scripts from leaking into the ResourceCache #109345). This seems to work fine, the ResourceCache can handle it and we can ensure that fully loaded scripts always have their path set correctly.

I retested a lot of MRP's with this PR + #113580

And so far those issues seem to still be fixed with this change:

• Validation error using Resource sub-classes and sub-resources with specific combination of preload/@onready/instantiation #95909
• Custom Resources that worked in 4.2.2 do not in 4.3 (does not recognize as resource?) #95789 -> Original MRP does not crash. The issue probably needs to be rechecked at some point because there is quite a lot of comments.
• Type casting may cause property access error with custom resource #93433
• Resource silently fails to load with specific combination of @export/preload/globals/script references #110394
• Resource loaded without being associated with script class name #87397
• Built-in script live reloading is broken again #99006 -> builtin output panel seems to be somewhat broken when printing every frame? But the console output shows that the reload does work.
• Script reference is broken when script is moved in FileSystem dock #101615 -> although this one always was a bit inconsistent for me