Added
* Add v4 API endpoint `job_start` to start job with `onDemandSparkCluster` option
* Add v4 API endpoint `job_stop`, `job_status` & `job_start_blocking`
Changed
* Started caching project_id
* Started using request session instead of creating new session every time
* Check if an App is running before attempting to stop it (in `app_unpublish`)