Explore adding a union type

I think this could leverage a lot of the same stuff the other wrapper types like List and Optional do.

Example from user:

from type import Union



### The reason why I'd like to have something like this is that I want to acts a combiner function that accepts both types of 
### values. I don't want to have to attach config parameters to descriminate the input and output type but it may the direction that I would need to go?
@solid(
    input_defs=[InputDefinition('provider_partials', List[Union[ProviderDataFrame, ProviderSeries]])],
    output_defs=[OutputDefinition(ProviderPartialDataFrame)],
)
def merge_provider_partials(context, provider_partials):
    context.log.info('Merging provider partials to providers')
    partial = pd.concat(provider_partials, axis=1)
    return partial

[kinghuang]

@schrockn Where can I learn about ScalarUnion? I can’t find anything about it on the docs site.

[natekupp]

For runtime types, we now have unions via PythonObjectDagsterType. Also looks like ScalarUnion is documented now, but @kinghuang please let us know if you need more info there