Bind missing default value for RichTextLabel methods.

[Daylily-Zeleen]

Bind missing default value for RichTextLabel methods:
void push_font()
void set_table_column_expand()

I found that void push_paragraph()'s default value of argument p_direction in c++ is Control::TEXT_DIRECTION_INHERITED, but use TextServer::DIRECTION_AUTO to bind.
I'm not sure this need to be fixed or not.

[akien-mga]

CC @RedworkDE - We should think about how we want to handle compatibility now that 4.1 is released.
We still run checks against the 4.0-stable API, we should start checking 4.1-stable (and see if we also still want to check 4.0, or only 4.1).

Edit: I wouldn't block this PR either way, if we add the 4.1-stable checks in a follow-up PR we can move the expected errors added here to the 4.1 file.

[RedworkDE]

This needs GDExtension compat methods a-la #76577, or this will break compat between 4.1 and 4.2 (we skipped that for 4.1 because of the more fundamental breaks) and then the validation error will go away as well.

[bruvzg]

This needs GDExtension compat methods a-la #76577, or this will break compat between 4.1 and 4.2 (we skipped that for 4.1 because of the more fundamental breaks) and then the validation error will go away as well.

Adding default values should not break anything, call arguments won't change. If it does, it probably should be fixed on godot-cpp side instead.

[RedworkDE]

The default values are part of the method hash tho, so a build looking for the old hash will no longer find these methods, which is what the compat methods then fix.

While it can be argued that the method hash should not depend on the default values, that is an entirely different issue and changing that is way beyond the scope of this PR (it would need to be done here in core tho)

[Daylily-Zeleen]

I think we should convert this pr to draft, and wait a pr to implement generate method hash without default values.
Or just bind compatible methods in this pr?

@RedworkDE , @bruvzg , @akien-mga what do you think?

[RedworkDE]

I would just add the compat methods here, as changing how the hash is computed comes with a whole bunch of complications, which adding the compat methods is pretty straight forward and quickly unblocks this PR.

[RedworkDE]

Seems fine to me

[YuriSizov]

Thanks!