Datasets not initally available when starting from scratch

[hjwilli]

When starting pennai completely from scratch, the only datasets initially available via the web interface are
(Adults, Breast Cancer, Gametes, Hypothyroid, Mushrooms, Readmissions); the initial datasets defined in pennai\lab\examples\Users\Users.json. It seems that after some time if the containers are restarted all of the datasets are available.

The cause may be that the process that does the initial loading of the datasets takes a while to run, and does so silently.

[hjwilli]

Discovered an issue... the 'npm install' command had only partially completed on machine (because I force exited the container during the initial build... whoops), meaning there was a machine/node_modules directory which was incomplete and the machine process was not starting correctly.

Updated the machine start script to run 'npm install' if it sees a machine/node_modules/.staging directory.

[hjwilli]

This seems to be stable as of fc25144.

Datasets are loaded during the startup process of machine instances, which register the datasets located in machine/datasets using the lab api (/api/v1/datasets/). Inconsistent behavior was caused by machine and lab starting at different times, machine instances now wait for lab to be up before attempting to register.