Initial create grid of chips

[andrewphilipsmith]

Essential

• [x] Loads the UK coastline from Strategi data.
• [x] Dissolves the coastline (multiline) into a single MultiPolygon feature.
• [x] Creates a geodataframe of polygons that are squares covering British National Grid, using the chip size in pixels (eg 32) and the pixel scale (10 metre).
• [x] Adds a boolean on_land column by intersecting the geodataframe with the coast polygon.
• [x] Save the geodataframe to a .parquet file.
• [x] Creates a raster that is geographically perfectly aligned with the geodataframe (i.e. one pixel of the raster == one row/geometry in the geodataframe).
• [x] Save the raster to a geotiff.
• [ ] Save the raster as an xarray.
• [x] Repeats all of the above for the "chips" and the GEE images.

Nice to have

• [ ] Configurable from the commandline. (_At present, there are a couple of config dicts at the top of pyveg/scripts/create_chip_grid.py that can be altered_)
• [ ] Add a column in chips_gdf with the ID on the parent image. (_This proved very slow to compute, though it could be made a lot more efficient. For now there are a couple of helper functions (get_image_id_from_chip and get_parent_image_id) in pyveg/src/grid_creator.py._)
• [ ] Link script into setup.py entry points.
• [ ] Define the output directory. (There is a param in the config dict but it doesn't do anything)

[andrewphilipsmith]

@crangelsmith If you have a chance to review this, that would be great. Feel free to merge it if you want to, or it can wait until I'm back.

• I am not sure what is causing the failing test or how important it is.
• If you do merge it, then please make an issue for the xarray task. Many thanks

[crangelsmith]

I've decided to delete the pyveg data analysis tests that are not at all relevant for us and use outputs from previous tests, which might change due to changes in the donwloading/processing modules. At the end we must only keep the tests that cover what we need for this project.