Use refactored PyCIEMSS lib

[fivegrant]

This PR refactors the service to use PyCIEMSS v0.1. The changes to the API are kept to a minimum but alterations are unavoidable since we are moving from "big box" to "small box" (atomic) operations.

PyCIEMSS features that still need to be incorporated into the service

• Pass parameters.dill into simulate (i.e. using inferred_parameters from calibrate for sample)
• Reimplement interventions (unclear which kind of interventions I should expose)
• Fix solution mappings for ensemble-simulate to work with the new format

Regressions that have now been fixed but need to be incorporated into service

• Progress hook with additional 'loss' field.

Regressions in PyCIEMSS

• Visualizations
• Progress hook
• Ensemble Calibrate
• Optimize operations
• Logging decorator missing from ensemble-sample (very minor)

[fivegrant]

Please note the following regressions

• solution_mappings don't apply in ensemble-simulate #52
• Visualizations are no longer provided #53
• ensemble-calibrate missing #54
• Replace missing optimize endpoints with a single optimize endpoint. #55
• API Container uses PyCIEMSS library #56

tl;dr: simulate and calibrate have been successfully refactored while there are various issues with everything else

Dashboard Results

simulate and calibrate now work for all except one models. This includes stockflow and regnets. All of the other other operations no longer pass