include variant phase in model

[kathsherratt]

see #102 and #123
Replacing location x time term with more specific term specifying a random effect for each dominant variant phase (identified for each location / week).

• prep-data.R : sources variant data from ECDC. Identifies dominant variant phase as the first week a variant became dominant in each country.
  • see variant-plot.Rmd for step by step code / visualisation of variant phase identification
• model-wis.R : replaces s(time, by = location, k = 40) with s(VariantPhase, bs = "re")
• model run (now reliably faster) and saved
• minor update to model plotting to plot effects of variant phase
• previous model and results archived