New scenario types and associated make_Rt methods

[SamuelBrand1]

This PR does two things:

• It creates an abstract type for Rt without renewal scenarios, and creates a set of concrete sub types aimed at setting the scenario behaviour (the scenarios are discussed in Flesh out multiple simulated scenarios vs one long one #52 ).
• It creates a set of make_Rt methods which determine the Rt time series set by the scenario along with docstring examples of how to plot the decision for visual inspection.

As well as extension of unit tests to cover the new structs and functions.

NB: I've taken the definition of "endemic" to mean a geometric mean of $R_t$ over the time period is unity,

$$\Big(\prod_{t=1}^n R_t \Big)^{1/n} = 1.$$

This approximately ensures a time average stationarity in cases, i.e. stable endemicity.

Closes #325 and #326 .

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.18%. Comparing base (849b707) to head (3275c6f).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #327   +/-   ##
=======================================
  Coverage   93.18%   93.18%           
=======================================
  Files          50       50           
  Lines         528      528           
=======================================
  Hits          492      492           
  Misses         36       36           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[seabbs]

I'd missed you were PR stacking so review comments on the other PR are mostly for here I think

[seabbs]

This all looks good to me.

[SamuelBrand1]

PR context

From f2f conversation @seabbs is a bit uncomfortable that we essentially have two patterns here:

• A type-dispatch type pattern for setting the "truth data" configuration (e.g. endemic vs non-endemic etc).
• A list of Dicts that specify the configurations of the inference models applied to each "truth data" model.

My argument is that because there are a small(ish) number of "truth data" scenarios we are considering, this is clearer/easier at the moment but we should consider a trait-based system to ship with an EpiAwarePipeline package (#229 ), for example have a trait for endemic truth scenarios which can have types added to it dynamically by the user to define a workflow (see SimpleTraits.jl for implementation).

NB: the truth data scenarios still contain a list of Dicts specifying the GI because the analysis extends to the effect of GI mis-specification.