Incorporate cadre allocation computations more into mainline

[celiot-IDM]

Phew ... this work started out with a small idea and turned into something very big.

The core work takes the cadre allocation computations that used to be done by functions in pace_post_process.R and moves them more into the mainline of process. So instead of using suite results CSV files as inputs, post-processing information can now be done immediately after an experiment finishes, based on the results structure returned by RunExperiments().

NOTE: THIS IS A BREAKING CHANGE. Old versions of the post-processing code won't work with the new Cadre allocation sheet format. A side-effect of making cadre roles user-configurable at the scenario/delivery-model level is that we can't naively combine suite result CSV files the way it was done before by ReadAndCollateSuiteResults(). (The key problem is that a user could configure a member ID - e.g "HEW1" - to mean one thing in one delivery model, but something different in another delivery model. Different delivery models have different cadre memberships - MergedModel = {HEW1 | HEW2 | FHP1 | FHP2}, but BasicModel = {HEW1 | HEW2 | FHP1 | FHP2 | RN | EHP} - so it's possible that in trying to create one big view that crosses several delivery models, we could end up aggregating things that shouldn't be aggregated.)

Annual overheads are calculated for all cadre members in all models, but I'm not sure how you want to access that information.

Here's a sample of the new approach.

library(pacehrh)
pacehrh::Trace(TRUE)

pacehrh::InitializePopulation()
pacehrh::InitializeScenarios()
pacehrh::InitializeStochasticParameters()
pacehrh::InitializeSeasonality()
pacehrh::InitializeCadreRoles()

pacehrh::SetGlobalStartEndYears(2020, 2050)
pacehrh::SetRoundingLaw("none")

results <-
  pacehrh::RunExperiments(scenarioName = "MergedModel",
                          trials = 20)

SR <- pacehrh::SaveExtendedSuiteResults(results, run = "Run_ID")
CA <- pacehrh::SaveCadreAllocations(SR)

summaryStats <- pacehrh::ComputeSummaryStats(SR, CA)

Two additional bits of magic ...

Magic 1: If you don't need to keep the output from SaveExtendedSuiteResults() (basically a much wider version of the standard output, but extended with Scenario and Task information), you can speed things up a bit by piping:

# Create the cadre allocation table without saving intermediate results locally or to disk
CA <-
  pacehrh::SaveExtendedSuiteResults(results) %>%
  pacehrh::SaveCadreAllocations()

Magic 2: Both SaveExtendedSuiteResults() and SaveCadreAllocations() take filepath parameters which will save their outputs to disk.

# Create CA table and save both the extended suite results and the cadre allocation results
CA <-
  pacehrh::SaveExtendedSuiteResults(results, filepath = "_SR.csv", run = "Run-100") %>%
  pacehrh::SaveCadreAllocations(filepath = "_CA.csv")

I have some more tidying up to do, but I'm going to move on to finish off the stochastic rate capping work now.

Addresses #51

[celiot-IDM]

@MeWu-IDM : Integration tests are failing, but I don't understand why. Over to you!

Update: Working now after merging input format changes from 244 into 51.

[MeWu-IDM]

does pacehrh:::SaveCadreAllocations not work with the results from pacehrh::ReadAndCollateSuiteResults after this change?
I ran into error: more than one scenario in suite results

[rhanIDM]

@celiot-IDM, is RoleID required to be unique for task time allocation to cadre to compute correctly?

[celiot-IDM]

@celiot-IDM, is RoleID required to be unique for task time allocation to cadre to compute correctly?

The overhead times calculation will work just fine with duplicate RoleID values. After that it's up to you (the researcher) to collate RoleID's with the role information embedded in the headers of the cadre task sheets.

[MeWu-IDM]

We were able to update the scripts to produce the same cadre allocation results (Run_simulation.R is updated)
Regression tests are also updated to cover the new cadre calculation.

[github-actions]

Package	Line Rate	Complexity	Health
pacehrh	90%	0	✔
Summary	90% (2445 / 2702)	0	✔

Minimum allowed line rate is 60%

[rhanIDM]

Run_simulations.R is updated to post-process new cadre allocations results structure. Model_inputs_demo.xlsx is updated to new structure and we were able to use these inputs to confirm the calculation results match the old results.