Fix typed arrays of scripts not being decoded properly

[mihe]

Related to #699.

This is likely the same issue as what's mentioned here, which seems to stem from the fact that decode_TypedArray doesn't take into account all the forms that typed arrays can be encoded in.

Typed arrays can conditionally encode either a uint32_t as the type, if the array contains a built-in type, or a String if the array contains scripts of some kind, and decode_TypedArray currently only accounts for the former and not the latter.

This PR fixes this issue by incorporating not only HEADER_DATA_FIELD_TYPED_ARRAY_BUILTIN, but also HEADER_DATA_FIELD_TYPED_ARRAY_CLASS_NAME and HEADER_DATA_FIELD_TYPED_ARRAY_SCRIPT, as found here.

Similar to what's already being done, this also just discards this new script name/path.

[DaelonSuzuka]

Hey, thanks for investigating and submitting a fix! Could you share some example GDScript that breaks the existing code so I can test this?

[mihe]

Oh, yes, sorry!

It's as simple as declaring a custom resource (or presumably any type really) as class_name MyClass extends Resource and then declare a typed array member variable somewhere else as Array[MyClass]. Once you hit a breakpoint where that typed array is in scope then you'll see the error mentioned here and the Locals, Members and Globals lists in the debugger view will all be empty.

[DaelonSuzuka]

Awesome, I can confirm that this fixed it:

Thanks @mihe!

[mihe]

Great to hear!

Any idea when there might be a new release cut with this included?