HITL - Handle episode lists instead of ranges.

[0mdc]

Motivation and Context

In previous iterations, the HITL server accepted episode index ranges in the format episodes: 100-104.

This approach had multiple issues:

• In a crowdsource context, it was not possible to "recycle" individual failed episodes. Entire ranges had to be discarded.
• If an error occurred during the middle of a session, data from the completed episodes was discarded.
• Special care was required to sequence episode sets:
  • Tutorial episodes needed to be copied every N indices.
  • Scenes required to be sequenced in order.
• Once an application was deployed, it was not possible to adjust task distribution.
• Ambiguous range boundaries causing human errors.

This changeset changes the episode format to accept ranges instead, in the format episodes: 100,101,102,103

How Has This Been Tested

Tested on distributed HITL application.

Types of changes

• [Development]

Checklist

• My code follows the code style of this project.
• I have updated the documentation if required.
• I have read the CONTRIBUTING document.
• I have completed my CLA (see CONTRIBUTING)
• I have added tests to cover my changes if required.

[eundersander]

See comments.

[eundersander]

Beware you are kinda duplicating the episode_filter logic in hitl_driver.py. Consider reusing that instead.

[0mdc]

This is a planned refactor post single-learn.

Lots of the code in the rearrange_v2 can be moved to habitat-hitl.

[eundersander]

LGTM!