Feature request: add ability to mock different api datasets

[mrshll1001]

It'd be good if users could mock different api datasets without having to remove their existing files. This could be used to mock an API with totally valid data, an API with totally broken data, and an API with mixed data etc etc.

This could be accompished by adding another layer of directories under data where each directory is named for an api dataset (a good word would be "profile", but we don't want to risk confusion with HSDS Profiles).

At the moment, the directory structure is as such:

data/
    services/
        example.json
    organizations/
        example.json
…

We could change it so that it is structured like this:

data/
    all-valid-data/
        services/
	    example.json
        organizations/
	    example.json
	…
    all-broken-data/
        services/
	    example.json
        organizations/
	    example.json
	…

And so forth.

Ideally, it'd be possible to pass an option to app.py which declared the name of the dataset directory to use. This way, a user can mock different API contents to test their program.