Add dry run flag to scope, so when launched we can check the args are valid.

[tomwilkie]

Fixes #1191, fixes #1559, fixes #1548

[jml]

Code changes are fine, but when I downloaded this branch and tried it out it didn't behave as I expected:

$ git checkout origin/check-args
$ make
...
$ ./scope launch --dry-run
b0ea1a0f2c481515e4eebfc781ee144d5561d61caa503e570f766c45bea5a655
Weave Scope is reachable at the following URL(s):
  * http://192.168.99.100:4040/
  * http://10.0.2.15:4040/
$ docker ps | grep b0ea
b0ea1a0f2c48        weaveworks/scope:latest                                            "/home/weave/entrypoi"   21 seconds ago      Up 21 seconds                                weavescope

I would have expected the docker container to terminate, and to not have been given a URL at which scope was reachable.

[tomwilkie]

Doing that you'll end up with a container stuck in a loop (scope exiting, the watchdog restarting it).

I could add a check to entrypoint.sh to block this, but entrypoint.sh has been nothing but trouble and one of the aims of this PR is to remove functionality from it. So as agreed I don't do that.

In the end I just documented that -dry-run is for internal use only.

[jml]

LGTM.