Dataset with splits

[BenGalewsky]

As a foundry dataset user I want to manage splits in a dataset so I don't have to download the entire dataset

Description

The DatasetCache class will need to manage splits alongside datasets.
The Dataset class will have a split_labels method to return the list of valid split labels.

Assumptions

1. The DatasetCache class' download_dataset method will be renamed download_dataset_split
2. The as_pandas accepts a split label argument.
3. split labels will be received from search metadata

Acceptance Criteria

f = Foundry()
datsets = f.search("DOI123/445")
labels = datasets[0].split_labels()
df = datasets[0].as_pandas(split=labels[0])

f = Foundry()
datsets = f.search("DOI123/445")
df = datasets[0].as_pandas(split="doesnotexist")