Forecast Import Speed Enhancements

This issue is to track the work necessary to enhance the current forecast import pipeline to support importing forecasts for the whole chain each night. Current measures extrapolated put us at ~12 hour to import forecasts for the whole chain, and also doesn't leave much room for emergency re-imports as needed. Still need to narrow down exactly what this enhancement will be, but the current recommendation is to explore removing the "zero" forecasts to significantly (up to 80%) reduce the amount of data sent to InFORM for a single forecast.

Summary

• Add Second Service Type to Forecast Import Files
• Remove all intervals that are all zeros
• Load Testing:
  • Run current forecast import load test for 2,200 stores
  • Run load test with second Service type added to the forecast import files
  • Run load test with second service type, and zero intervals removed
• Regression test using forecast imports without zeros