consider gutting some commands for maintainability

[ilaflott]

From epmt --help:

    source              Enable instrumentation of subsequent shell commands
    start               Begin EPMT job data collection
    stage               Generate job archive
    annotate            Annotate a job
    dump                Print job metadata from archive or database
    schema              Dump the database schema
    migrate             Migrate the database schema
    convert             Convert csv from v1 to v2 format
    run                 Execute a process
    stop                Stop EPMT job data collection
    daemon              Run the EPMT daemon
    shell               Start an interactive IPython shell, see 'epmt python'
    python              Run a Python script under the python interpreter
    gui                 Run the EPMT dashboard GUI
    unittest            Run EPMT unit test suite
    integration         Run EPMT integration tests
    drop                Drop the entire database, Caution!! Irreversible!
    retire              Delete jobs/models, retention policy in settings.py
    submit              Submit/ingest job into database
    dbsize              Find detailed size of database
    check               Verify EPMT Installation
    delete              Delete jobs from the database
    list                Display all jobs in the database
    notebook            Start iPython Notebook environment
    explore             Explore a particular experiment, looking for outliers
    help                Display help with extended configuration information

so epmt is expected to-

• data-scraping/annotations
• data-scraping control
• database maintenance, creation, migration
• provide access to ipython and python
• run it's own unit AND integration tests
• provide graphical interfaces via jupyter and visual dashboards ala grafana
• data analysis functionalities

In an era of microservices, this makes epmt surprisingly monolithic

[ilaflott]

partially addressed by #53