OPS-1593 As an operator, I want sudo_user converted to become user in roles and plays, for compatibility with Ansible 2.x

OPS-1593 As an operator, I want sudo_user converted to become user in roles and plays, for compatibility with Ansible 2.x

OPS-1593 As an operator, I want sudo_user converted to become user in roles and plays, for compatibility with Ansible 2.x

Ansible turned 0640 into 420 and 0750 into 488, only one of which are
valid file modes.  Quotes prevent it from doing so too early.
Fixes d2f1595c.

During the edxapp refactor, three tasks were combined, but the modes
were all squished to 0644

Improve docs.

This will connect to a cluster, find the hidden secondaries and upgrade
them to the current version specified in the mongo_3_0 play.

This required the creation of a new library module since our two current
ones return rs.status and call replSetReconfig, but neither return
rs.config, and rs.config is what tells you if a host is hidden and the
priority, which are needed for this.

http://docs.mongodb.org/manual/release-notes/3.0

SERVER-22970  Background index build produces an index with mismatched
index keys and documents

We were implicitly relying on mongo to return the list of members in
their _id order, which is absurd and I have no idea how we didn't catch
it previously.

We could do a little list comprehension to pull out the member with the
same _id,  but it seemed easier to just sort both lists and retain the
original intent.

I removed this in 9ae2caa0 so it's a no-op

Updated ansible config to increase ControlPersist (fix timeouts on long running tasks for remote hosts)

This appears to be the only way to ask it to look in multiple regions
since it doesn't support AWS_REGION or --region.
Alternately, we could start having independent .ini files for deploys.

* Makes the mysql role a dependency of edxlocal
* Moves memcached install into new role
* Installs memcached if EDXAPP_MEMCACHE contains localhost

It is part of the default AWS AMI though, which is why we haven't
noticed.

When running on a dirty image (such as a sandbox base image or an AMI
build) this clears static files so that collectstatic will copy
everything.

Django's collectstatic copies files and doesn't preserve timestamps,
then only copies "newer modified files".  This leads to a condition
where old static files are in /edx/var/edxapp/staticfiles with a
timestamp of Yesterday, and we add a new dep which comes down from npm
with a timestamp of last week.  Collectstatic will not copy those.