it was becoming too complex; now we are running multiple containers
from the same image. Coordination between those processes is not needed
(db is where it happens), so it should work.
waiting psql to be alive is a duty moved to a wrapper script "wait-db"
reasons:
1. nginx resolve names at startup. This requires strong dependency
between containers. Strong dependency is bad.
2. apache is used elsewhere, so less knowledge needed to develop the
whole stack
a syslog image will log everything it receives on stdout; so every
service that has a hard time logging can just send everything to it.
apache/ttrss configured to do it.
also, state dirs are moved to /var/cache/ and /var/lock/, thus avoiding
their unneeded "copy" to host.
ICONS_DIR is still pointing to documentroot, so this is not yet
complete.
