Implement React-based document capture selfie liveness step

[aduth]

Previously: #3999

Why: As part of the React-based document capture, the user will be prompted to take a photo of themselves for liveness verification.

The primary purpose of this pull request is to begin flesh out a common input component which can be used across ID documents capture and selfie capture for uploading an image or direct capture via the Acuant SDK.

It implements:

• The text and photo input of the Selfie step
• Device capture detection:
  • Displaying a "Take photo" button if the browser supports capture
  • Redirecting a click on the photo input to use Acuant capture if supported. Otherwise falling back to retaining previous behavior to show a file input.
• Updating file input to support both file values (from file input) and data URLs (from Acuant)

It does not implement:

• Triggering Acuant to capture in selfie mode
• Error handling when the Acuant SDK responds to capture with a failure (as is the case when using Chrome to simulate)
• Per implemented note about supporting both file value and data URL string value, after discussion with @ironman5366 and @jmhooper , this will be simplified in the future to just support string values
• Additional upload options for mobile devices to choose to upload a file even if capture is also possible

Screenshots:

Mobile	Desktop

[aduth]

Note: The pull request at #4012 complements this in that the changes there may simplify what is required here after a rebase. As such, it may be easier to review #4009 before this one, as there are fewer changes.

This has been rebased after the merge of #4012, reducing overall changes slightly (from +326/-81 to +291/-77).

[zachmargolis]

LGTM