Move a number of scripts which were managed in a separate repo
into management commands.  This way they can be more easily executed
either manually or by cron jobs.

Settings have been reorganized and how different parameters are
determined has been reworked to be compatible with both a Puppet-based
VM deployment and a containerized deployment

Add some information to VOEvents and rearrange some things:
  * Add EM Bright information (BNS, NSBH, BBH, Terrestrial) in a
    Group in the What section
  * Put ProbHasNS and ProbHasRemnant in a Group and rename to 'HasNS'
    and 'HasRemnant'

See https://git.ligo.org/lscsoft/gracedb/issues/77

Change the Retraction field to be an int (rather than a bool)
to match other "flag-type" parameters.

Since we only have one API endpoint now, we only need one skymap
URL in a VOEvent, rather than three.

Also removes need for pylal, which is no longer available.

Fixing two migrations
  - ligoauth 0004: properly delete User instance for gstlal-spiir-gpu,
    not just LocalUser/RobotUser
  - ligoauth 0013: deleted, since gstlal-spiir-gpu has been deleted,
    we don't need to create a RobotUser instance for it

Restrict access to RSS feeds to internal users