fail better when task download timeouts

[ripienaar]

When a download timeout you'll get this:

$ mco tasks run package --action=status --name=puppet-agent -v
Retrieving task metadata for task package from the Puppet Server
Discovering hosts using the choria method .... 2
Attempting to download and run task package on 2 nodes

Downloading and verifying 1 file(s) from the Puppet Server to all nodes: ▘  1 / 2
warn 2018/03/28 03:53:25: client.rb:279:in `rescue in start_receiver' Could not receive all responses. Did not receive responses from dev-puppetserver.example.net


No response from:

    dev-puppetserver.example.net

Could not download the task package onto all nodes

realistically though, if it does not download a file in 15 seconds its not going to happen. We could timeout wrap the http get and fail better - rather than just not responding on timeout.