Expose dt/steps_per_day control

[richfitz]

Just the core parameter handling part done here. Someone needs to check that the epi looks ok.

We should expand the testing to cover things like Rt but I wonder if this might actually not be that bad

Fixes #187

[codecov]

Codecov Report

Merging #188 (2de2a6b) into master (56c879b) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #188   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           14        14           
  Lines         1196      1200    +4     
=========================================
+ Hits          1196      1200    +4     

Impacted Files	Coverage Δ
R/carehomes.R	100.00% <100.00%> (ø)
R/parameters.R	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 56c879b...9634e39. Read the comment docs.

[annecori]

I added some tests but the Rt test is failing:

sircovid/tests/testthat/test-carehomes-rt.R

Line 147 in 50e118c

test_that("Can compute Rt with larger timestep", {

Maybe @edknock you could look at this?
I now have the same model with dt = 1, 1/4 and 1/100 and the Rt are all different, although not to the extent that it could just be a forgotten *dt or /dt somewhere. Still some noticeable differences (5-15%) which require some investigation...

[annecori]

Update on the above. I implemented a function to update the gamma parameters before running the model, so that the resulting mean duration of the compartments using the discretised gamma would match the mean of the continuous gamma distribution. This has fixed the Rt calculations - which are now similar across different dt values, but generates substantially different epidemic dynamics - e.g. attack rate is very different with different rt values.
I think this means we need to rethink the calculation of Rt - I'm not sure it is correct to calculate the Rt using the discretised mean duration - maybe we should just use the mean of the continuous distribution? @edknock would be great if you can have a think about this.