Better typing and checks for index header

[AKuederle]

We should allow users to clearly specify what header values they expect.
This could be used to actively check if get_index is implemented correctly and returns the expected values and it could be used to improve the typing of label and group.

To do that we could allow the user to specify a typing.NamedTuple

class MyDataset(Dataset):
    _INDEX_HEADER = NamedTuple("MyDataset", participant=str, test=str)

Notes:

• If we use this as return value for group, this will not be correct typing, as group likely only returns a tuple with a subset of these fields. However, I think this is still better than the current typing which is Tuple[str, ...]. At least it allows the IDE to autocomplete the subfields
• We could set _INDEX_HEADER as default to Tuple[str, ...]
• When checking the index during creatiion, we should check that our tuple type has the same name as the dataset (otherwise the printed output might be confusing.
• We should further check of the df returned by get_index has the correct column names AND the correct types per column

[AKuederle]

This is partially implemented in #113.

This PR does not yet add a check if the Index matches the header