Add support to create different configs with `push_to_hub` (+ inferring configs from directories with package managers?)

Now one can push only different splits within one default config of a dataset.

Would be nice to allow something like:
```
ds.push_to_hub(repo_name, config=config_name)
```

I'm not sure, but this will probably require changes in `data_files.py` patterns. If so, it would also allow to create different configs for packaged modules datasets.