Allow audio subjects in the classifier

[lcjohnso]

As a researcher,
When I have research subjects captured on audeo,
I should be able to put my project on the mobile app,
So that mobile-based volunteers can help me do my research.

Here is a screenshot/diagram of what this should look like: (please put the screenshot here rather than a link to an inVision, because open source contributors do not have access to our private inVision account)

Here is a project we can test this on: @lcjohnso please let me know if we have such a project in prod, and/or let's make one in the "mobile app all workflows" sample project on staging!

Additional context

Currently, all subjects are rendered through the react native Image component. In order to open more projects to the app, we should find a way to render video subjects.

We currently have X number of projects in production that could take advantage of this feature.

[eatyourgreens]

I'd be very interested to see a workflow that works entirely with VoiceOver on iOS, so classifications can be made without needing to see the phone screen at all.