Input type BfabricResourceDataset extensions

[leoschwarz]

1. Input type BfabricResourceDataset should detect the Resource column by type
   We currently only allow detection by index or name.
   This is OK for dataset templates.

    column: int | str = "Resource"
    """Column name or index containing the resource IDs. (case insensitive if string)"""

But since we have typing information about the column available, it would be preferable to use it for greater compatibility. The current approach also does not allow defining fallbacks etc.

The main challenge here is finding a simple design which allows enough freedom but does not become too complex either, which is something I still have to think about a bit.

2. We should probably allow for legacy datasets

[leoschwarz]

One way would be to have strategies:

strategy: Literal["by_name", "by_index", "by_name_or_type"]

and then have conditionally defined fields depending on which of these is chosen. The downside would be that this is not very explicitly modeled in the schema.

Using the name would be perfectly acceptable if dataset templates are used.