Stanford datadog playbook updates

fixing handler name

Jarv/discern supervisor

Some general cleanup of the edxapp role, adding certs to continuous integration

Move xserver to supervisor

Fixes for edxapp: Permissions & supervisor startup/configuration

`supervisorctl reload` not only reloads the configuration files, but
also restarts all services, which causes all services to be reloaded
several times, and causes errors when the playbook later attempts to
start the services after changes have been applied.

Right after they are added for the first time, the commands are not available
until the groups are added.

For pip packages which create group-restricted venv files (httplib2).

Update worker server wait condition.

Fix rds ingress groups.

Jarv/cert fixes

Updated Vagrant for refactored roles

supervisor-ora: Fix inverted xqueue login/password in ora configuration

edxapp-devpi-fix: Add dependency on supervisor & fix venv install

of rbenv and virtualenv roles

Jarv/gather assets fix

Introduce POST size limit for ORA submissions

This was causing problems when gather_assets ran

The ORA machinery gets gummed up when it gets very large submissions and
throws a lot of 500s.

Co-Authored-By: Jason Bau <jbau@stanford.edu>