Add UK data from gov.uk api

[kathsherratt]

• uk_api.R sources all UK data from the coronavirus.data.gov.uk service.
  • The heavy lifting code is taken from their developers guide
    • we should flag to them and/or add as author
  • Data are left in raw state to highlight different data reporting (varying dates by test/publish/death) and the range of new variables available (testing, hospital data)
• get_regional_data.R has special UK switch to avoid the raw UK data going through the standardised formatting pipeline
  • very open to ideas on better balancing returning all data vs standardised clean data
• README, DESCRIPTION updated

This PR probably replaces #51 (which gets hospital data only) - but many thanks to @qleclerc for kicking it all off!

[qleclerc]

Thanks Kath! I'll rely on that for future data updates on my end instead of my first script in #51 , thanks for setting up the interaction with the API :)

[kathsherratt]

This is now good to go pending review @seabbs.

Latest commit details:

UK API now works properly and is integrated to return data as with other countries in the package.

• Own internal function to interact with API, so no dependence on the PHE R package ukcovid19
• Data are returned clean (with standardised variable names, as with other countries) but alongside all the raw variables
• The old UK code is still here but is commented out. Suggest we either
  • delete,
  • move to a back-up folder in case the new code breaks (!)
  • pull out any useful bits and store separately - as it has some useful tables of lower level region authority codes

See also #61 with tasks that this commit completes.

Outstanding issues (no action required yet):

• England regional deaths data are returned as NA. This is as it comes from the gov.uk dashboard, but this is expected to change when data are added soon. This will automatically get returned whenever it is added. See issue here
• England regional cases data may "soon" be available by PublishDate. This will not break code, but when this comes through we should make a small tweak to use this consistently as the cases_new metric.

[seabbs]

This looks good. I am not really clear what is being lost with the old data estimates? Can we still map the new data easily? If not is there anything we can do about it? If nothing is added suggest drop for now (code will still be in the commits if we need to go back and get it).

[seabbs]

It looks like all the UK tests are currently failing. These probably just need to be gone through and have their expectations updated.

[kathsherratt]

Nothing particularly lost by removing the old code, if it's in the commits then that's fine as a back-up.

I am re-writing the tests.

[seabbs]

Made a few edits (changing file name + fixing merge issues)