Cookbook TODOs

[bgoerdt]

1. Reprocessing data
2. Recovering from job failures and delays
3. Consuming event/change records into current state
4. Support "point in time" queries via Partitioning
5. Data modeling as a denormalized table of many columns
6. Removing sensitive values from datasets (PII)
7. Alerting on Job failures
8. Cleaning up Spark staging files
9. Controlling the number of output files
10. Creating a single file with a predictable name
11. Improving Job Performance
12. Only processing un-processed data