Tweet strategy/data issues

[mattkerlogue]

README now contains an outline tweet strategy as well as details on data availability/issues for each tweet. In summary, potential strategy is:

1. Tweet starting location with street map of the location
2. Tweet stating nearest public transport facilities
3. Tweet aerial photo of the current location
4. Tweet nearest bathing water quality information
5. Tweet some key statistics about the local area (population, age profile, employment rate)
6. Tweet nearest RNLI lifeboat
7. Tweet flickr photo
8. Tweet nearest non-GB location.
9. Tweet local MP/MSP/MS
10. Tweet distance from starting position

Outstanding matters to consider/decide:

• Process data at "start" of walk and prep tweets to be sent, or run relevant code for each tweet on each execution (former reduces computational load)
• Likely need to build dataset for Scottish bathing water and Senedd members, should dataset for other 'constants' also be made (e.g. all GB bathing water, politicians)
• Use Mapbox or Open Ordnance Survey data for maps?
• Use Mapbox navigation API for distance to other points (e.g. walking distance to train station/bus stop)
• Statistics/politicians requires lookup conversion from long/lat to administrative geographies. NOMIS API has experimental long/lat lookup service, or should repo contain boundary data (from ONS Geoportal) and do local look-up.
• RNLI data also includes historic data about lifeboat launches, might be interesting to compute stats
• Need to decide on reverse geocode service for non-GB location, potentially map these two points? e.g. can draw simple map using {ggplot}
• What delay between tweets (e.g. one walk per day or multiple walks)? Can/should {rtweet} post subsequent tweets as a reply to the previous tweet?
• What distance limit should be placed on Flickr search? Should photos be scored as per narrowbotR? What to do if no photo found, should bot skip to next tweet, not tweet or make a comment about forgetting its camera?

[mattkerlogue]

As discovered in 3aae8d0 DfT NaPTAN data is messy - best to build a reference dataset.

[mattkerlogue]

Where to get tide data from? (Could I compute my own tables????)