Add variant phase

[kathsherratt]

Adds infrastructure to include effect of variant phase. Variant phases in each location identified by the week from which a given variant became dominant. Addresses #137 / #102

Data processing

• data/variants/: added csv files with raw data on variants % for 30 countries (ECDC) + Switzerland + UK
• data/README.Rmd: documents sources for raw data
• R/utils-variants.R: Variant phases identified by the week from which a given variant became dominant.
  • Currently this code implements a hacky version that takes the median date of variant dominance across all countries.
  • This should be replaced with country-specific dates for each variant phase. Code to automate this exists here, but is currently commented out. There are some countries which need manual review/adjustment as variant phases don't follow the expected pattern.
• R/process-data.R: Variant phase added to data pipeline

Main results

• R/analysis-model.R
  • Model updated to include effect of variant phase, replacing model term for location x time
  • Added univariate models for each factor in joint model, to allow comparison for all factors in supplement

Supplement

• Adds a plot of variant phases per country
• Adds a plot of the partial effect of each confounding factor on log WIS to the supplement

Scope creep

• location variable capitalised to follow convention of naming modelled factors in sentence case (R/analysis-descriptive.R)
• minor cleaning up comments and unused code (R/process-data.R)

To do:

• Update text in methods
• Update text in results
• Replace code to give variant dates per location