Simplify installations about fireq HOT 1 CLOSED

Next things have been done for this.

Combine build and deploy steps

In general there is still separate build and deploy steps, but grunt build is now running during deploy step, so all env variables are available already, so it's also kind of simplification, developers often use env variables in superdesk-client-core and liveblog too, so I don't need to keep and modify config.json file in fireq repository. Should be easy to maintain fireq.

Get rid of supervisord

So now we have systemd service file with honcho inside, looks better, also it's possible to create another systemd service for testing purposes (like running e2e tests).

Use only settings.py for server configuration

I gave up this idea, because we have too many different projects with different settings.py and moving them to fireq repository is just crazy. Anyway it still needs some env variables declared somewhere, for client SUPERDESK_URL, SUPERDESK_WS_URL are needed, PYTHONUNBUFFERED=1 and C_FORCE_ROOT=1 for honcho, etc. So for simplification (from user perspective) we now have one /opt/superdesk/activate.sh where we load virtualenv and declare some env variables, there is optional /opt/superdesk/env.sh which uses in CI and loads in activate.sh script.

There is also new build&deploy templates tpl/superdesk-new without main superdesk repo and with minimal set of files to run superdesk using just cores. It was created mainly for inspiration in regards of #8 (rethink main repo), union /opt/superdesk/activate.sh script came from this attempt... It can be used for user installations in future.

from fireq.