Improve use of geometries in ICES station dictionary

[RobFryer]

Need to tidy up the way in which we work with station geometries in the ICES station dictionary

Currently the station dictionary has nominal co-ordinates and co-ordinate ranges as well as a geometry collection. The nominal co-ordinates and ranges are created from the geometry in the station dictionary download, but can lead to inconsistencies due to rounding / reporting errors.

Also, the nominal co-ordinates do not necessarily lie within the station polygon.

Also have some weird inconsistencies with station 13771 - see #544 and #545

Suggested improvements:

• if nominal co-ordinates explicitly provided by submitters, check they lie within the polygon
• if not explicitly provided, generate suitable internal co-ordinates
• deprecate nominal co-ordinates and ranges from download, and pick up from geometry
• deal with special cases such as 13771

Need to co-ordinate with a package release, particularly for the deprecation of nominal co-ordinates and ranges

[HansMJ]

@RobFryer I have made a related ticket on this is the ICES station dictionary ticket system, but I am not sure if this is visible from outside. The suggested action is to always add a point to a station geometry when they are being defined. The point should always be inside the station polygon. This suggestion needs to be approved for all uses of station dictionary. Alternatively we could default the point for a polygon to 'st_point_on_surface' which should always be inside polygon.

• Currently when a polygon is drawn, there is no point saved in the geometry collection
• If a station is created as a point with lat/lon ranges, the central point is saved in geometry collection

In the XHAT station download the point is

• Taken from geometry if is there
• If not then point is taken as the centroid of bounding box
• Decimals (point and lat/lon ranges) are limited to 5 (~1m precision). This could be increased