Investigation showed that the high rate of cases in the UK in January
caused a significant discretisation error between the chain binomial
model and continuous time analogues.  The previous Rt calculation
made use of a continuous time approximation, and so gave biased
estimates of Rt.

The new implementation calculates Rt using expectations of the chain
binomial model in terms of transition probabilities, not rates as
previously.

Fix for weekday effect in prediction

See merge request !37

Previously, when generating forward predictions, the model used the latest value of
gamma1 for all time points.  This created a prediction artefact depending on what the
last day of the observed timeseries was.   This is now corrected by projecting the weekday
forward in time.

Fix to trap ChunkedEncodingError in data download

See merge request !36

Fix download retry

See merge request !35

Add option to push to AWS buckets

See merge request !34

CHANGES:

1. Add option to upload results to AWS S3 bucket
2. Remove generation of insample plots from pipeline

Modified Rt calculation

See merge request !33

CHANGES:

1. The posterior NGM is now not saved, only the sum over all destinations to give $R_{it}$
2. Replace Rt as the dominant eigenvalue of the NGM with weighted average by population size.
3. ngm.nc now contains both $R_{it}$ and $R_t$ posterior predictive values.

CHANGES:

1. `thin_samples['init_state']` --> `thin_samples['initial_state']`
2. `thin_samples['events']` --> `thin_samples['seir']`
3. add time dimension to next generation matrix
4. update downstream scripts to changes.

This reverts commit 53f6937b

Regular TF graph mode appears to be faster for the current code
implementation.

Adjusted pipeline to ensure config.yaml is written to results dir first.

See merge request !32

Fix geopackage geometry joining issue

See merge request !31

Use xarray/netCDF4 throughout the pipeline

See merge request !30

CHANGES

1. Most tasks touched to use netCDF4 inferencedata file.
2. Added netCDF4 metadata copy to covid.util
3. Updated model_spec.py to use xarray
4. Updated covid.ruffus_pipeline to use netCDF4 files, adding pipeline metadata
5. Updated pyproject.toml
6. Removed chunk specification in inference.py.  Defer to automatic chunking for now.

Added cumulative incidence to `summary_longformat.xlsx`

See merge request !29

CHANGES:

1.  Added code in `summary_longformat.py` to calculate cumulative incidence;
2.  Switched from `openpyxl` to `xlsxwriter` for lower memory usage when writing
    to summary_longformat.xlsx.

Added daily absolute incidence and incidence per 100k to summary_longformat.py

See merge request !28

Added insample7 summary to summary_longformat.xlsx

See merge request !27