Make some Image methods static

[KoBeWi]

Follow-up to #60739 (comment)

[akien-mga]

I think this warrants further refactoring, it's weird to have static create_empty and create_from_data and non-static create with two variants.

Maybe:

• Move create(int p_width, int p_height, bool p_use_mipmaps, Format p_format) logic to create_empty (and rename to create to match bindings)
  • Or keep the create_empty name and use that for the bindings too. We should strive to have consistent between core C++ API and bindings for GDExtension's sake.
• Move create(int p_width, int p_height, bool p_use_mipmaps, Format p_format, const Vector<uint8_t> &p_data) logic to set_data and use it in create_from_data.

[reduz]

I misunderstood what you were doing. I am not entirely sure its good to make these static, since they are operating on the image itself.

Changed my mind

[akien-mga]

I misunderstood what you were doing. I am not entirely sure its good to make these static, since they are operating on the image itself.

It's only the create methods which are static, I think that makes sense? It creates and give you a new Image.

[KoBeWi]

I think the internal create() method could be renamed to create_data(), because that's what it actually does.

[KoBeWi]

I renamed the internal create() to initialize_data() (I didn't use create_data() to avoid confusion with create_from_data()) and also replaced remaining old calls with static methods where applicable.

[akien-mga]

We discussed this in a PR meeting.

The change seems good, but we're concerned that it will lead to more problematic situations where users are unaware that their code is no longer working when using the previous format (which would still work silently).

We should find a proper solution to this UX issue in GDScript before merging more such changes that will impact users of the beta version.

[akien-mga]

Thanks!