Merge coincident pole and antimeridian nodes in structured grids

[rajeeja]

Closes #1689

• _read_structured_grid matched nodes in the lon/lat plane, so the nlon+1 nodes at each pole and the two sides of the antimeridian were never merged.
• Now matches in Cartesian space, so nodes that are the same point on the sphere collapse: 899 nodes at 1 deg, 3599 at 0.25 deg.
• Merging leaves each polar quad with a repeated corner, so those faces are now stored as triangles padded with the fill value.
• Pole valence is nlon as expected (36 at 10 deg, 360 at 1 deg) and V - E + F == 2 confirms a closed sphere.
• Regional grids that touch neither pole nor the seam are unchanged.
• Two regression tests added; the global one fails on main.

[rajeeja]

Checked the docs and user guide for behavior changes: both datasets in user-guide/structured.ipynb are unaffected (air_temperature is regional, ersstv5 spans 0-358 deg so the seam never closes), and neither collapses a single node. Also fixed tol to stay in degrees rather than silently becoming a chord length, so the documented parameter keeps its meaning. Only grids that actually touch a pole or the antimeridian change.