Add support for running inside Docker containers

[marmistrz]

This change takes advantage of the Docker capabilities present in golem-unlimited. Closes #12. Closes #2. Closes #29.

Changes:

• a new, slimmed down Docker image has been introduced and pushed to the Docker Hub.
• the directory structure has been sanitized (/app/, /output/, /input/) and a default location for the output directory (/output) has been added
• we no longer require the administrators to take care of the tools that need to be present on the provider machine or the inter-node connectivity - this is now automated thanks to use of Docker.

TODO:

• [x] properly set the user (currently we run as root) [needs-gu-changes]
• [x] set the working directory to output [needs-gu-changes]
• [x] set the user to mpirun [needs-gu-changes]
• [x] fix the output directory in make/cmake (they're different, I've commented out the mv_cmd)
• [x] automatically start/stop the container [needs-gu-changes] (now it's a mess)
• [x] do proper separation of input, output & app data. [partially done]
• [x] simplify the tmp-related code
• [x] check what happens if an executable doesn't exist inside the container?
• [x] add extra logging to gu-client
• [x] working deploy+exec
• [x] deploy the ssh keys
• [x] test on single node
• [x] test on multiple nodes, with different OSes
• [x] update input&output
• [x] check if we're still affected by the mca btl hack (nope)
• [x] upload input to all the nodes
• [x] add a default for output
• [x] enable SYS_PTRACE capability

TODO-GU:

• [x] Autostart of Docker containers needs to be merged: golemfactory/golem-unlimited#241
• [x] ExecOptions PR has to be merged: golemfactory/golem-unlimited#243
• [x] cap_add PR has to be merged: golemfactory/golem-unlimited#253

GU blockers:

• [x] Automatically closing Docker containers issue has to be implemented by the GU team. golemfactory/golem-unlimited#246
• [x] Command::Exec doesn't return proper errors in case the program execution fails under dockerman: https://github.com/golemfactory/golem-unlimited/issues/250

Deferred issues: (follow-up issues will be created)

• for efficient shared memory in OpenMPI we may need bigger /dev/shm, Docker limits it to 64M. Probably using --tmpfs /dev/shm should be enough to guarantee unlimited /dev/shm
• handle GU processing error at one place.

[marmistrz]

The PR is basically ready, we're only waiting for the GU blockers.

[marmistrz]

We'll probably need to have --cap-add=SYS_PTRACE for MPI to work under Docker efficiently. https://github.com/open-mpi/ompi/issues/4948

[marmistrz]

Because this PR was pending for a very long time, I'll merge this PR without a review as an exception. @kubkon, please take a look when you have a while, I'll address the issues (if any) in a subsequent PR.