* Do not automatically kick off all jobs that yet have an entry in wh_etl_job_schedule table.
Instead simply schedule an entry and wait for its time to come.
* Make the ETL actor pool size configurable and set the default to something more conservative.
* Major refactoring
- Move job-spcific properties from wh_etl_job_property table to .job files
- Use the job file name instead of numeric IDs to identify ETL jobs
- Use reflection to create ETL job class at run time instead of relying hard-coded enums
- Drop ETL job-related APIs as they're no longer needed
- Drop wh_etl_job, wh_etl_job_property, wh_etl_job_execution tables
- Add wh_etl_schedule & wh_etl_history tables
* Rename web to wherehows-api and update README.
* Rename backend-service to wherehows-backend
* Rename metadata-etl to wherehows-etl
* Rename hadoop-dataset-extractor-standalone to wherehows-hadoop
