[Penpot] Import individual token type files as a single token set

### Is your feature request related to a problem? Please describe. I tried splitting a token set into individual files for readability. Right now our color set contains a lot of definitions, which normally makes the whole file span over 3k lines. I divided it into: ``` base/ colors.json spacing.json ... ``` I tried playing with metadata in all kinds of ways, but I couldn't import it into a single set called `base`. Instead, they all appeared in the group `base` as individual sets for color, spacing etc. Each of those contained only a single set of token types (ofc. depends if I correctly set all tokens to a single type in the source files): <img width="447" height="810" alt="Image" src="https://github.com/user-attachments/assets/d07289d0-daac-47ce-9c1c-34f59fe722b6" /> As you can see, there is `base/fontWeight` which only contains font-weight tokens, similarly colors define only colors etc. I have tried: - omitting root set name at all (e.g. `color` key from root object in c