Request: Circulair units (wrap-around for grid/coordinate domain, or time-units)

[jipclaassens]

Could be part of the solution for #721 But also handy for time-units as pointed out by @cjacobscrisioni

[MaartenHilferink]

For circular 1d domains (such as timecycles), existing operations can offer some proper results, such as:

• modulo (%)
• dist(a, b) := min_elem((a-b) % size), (b-a) % size).

For 2d cylindrical, torus, or latlong-sphere raster topology, specific attention is required for:

• potential
• diversity
• districting
• griddist

For 2d vector data on cylindrical, torus or spheres require adjustment of:

• area
• arc_length
• connect
• connect_info
• dist.

[MaartenHilferink]

The following adjustments can be made with existing operations in order to account for a/the East-West projection boundary,(thus not with the poles):

• focal statistics on lat/long: define: third_rel_s := first_rel+second_rel; third_rel := point_xy(pointcol(third_rel_s)) % pointcol(Bounds(domain_grid)), point, pointrow(third_rel_s)); // check that -1 % 40000 == 39999
• focal statistics on Mollweide: identify for each row with West column should be considered as East and vice versa, and apply in adjusting third_rel_s.
• Districting on lat/long: districgt East and West maps separately, abnd then compare stitched together boundary-cells to identify common districts; then renumber using the result of connected_parts.
• Districting on Mollweide data: same, but stich using the jumping East West boundary.
• Potential: embed in larger study area and cut or add results

The following operators require some adjustment to handle a stitched projection boundary:

• griddist for LatLong
• Potential for LatLong

I do not see feasibility for projection aware operations on mollweide data for:

• Potential
• griddist