Create job manager, enable append tables, add metadata tables (#14, #33)

[ssciolla]

This PR aims to enable multiple jobs (beyond the first, inventory.py) by creating a run_jobs.py as the Docker start point. Loading and management of environment variables has also been extracted to its own file (environ.py) so it could be loaded as needed by other parts of the application. run_jobs.py also records job history by updating two new database tables, job_run and data_source_status. Records should not be dropped from these tables, so I also introduced an "append table" concept, allowing some tables to be preserved by specifying them in env.json. Some additional details are provided below in the task list. This PR aims to resolve issues #14 and #33.

• Move inventory.py and gql_queries.py into canvas and rename to course_inventory.
• Enable what I'm calling "Append Tables" by updating db/db_creator.py.
• Extract env.json loading from various places and create environ.py
• Implement Job and JobManager classes in a new run_jobs.py file
• Create job_run and data_source_status tables for storing job metadata
• Add metadata validation and creation to run_jobs.py, including creation of ValidJobName and ValidDataSourceName enums.
• Update course_inventory.py/inventory.py to use APPEND_TABLE_NAMES and environ.py, and to return data source dictionaries that job_run uses to create data_source_status database records.
• Update README with a generalized overview, new variables, and a section describing how to implement a new job.

[ssciolla]

I've opened this for review. I will work on adding a section to the README providing instructions on implementing a new job.

[jonespm]

I think this looks good! We could improve the sleep loop and the append tables, but we could also fix those in a future issue too so approving.

[lsloan]

I've suggested some minor changes and asked questions, but overall I like it. Looking forward to using some of these features with the MiVideo code.