Resolve custom-labelled dimensions by name in GeoPandasInterface

[gaoflow]

Fixes #840.

A dim('status') expression builds a Dimension whose label is the column name. Dimension.__eq__ compares labels, so when the element declares that dimension with a custom label — vdims=[('status', 'Conservation Status')] — dataset.get_dimension(dim) returns None, and GeoPandasInterface.isscalar crashes with AttributeError: 'NoneType' object has no attribute 'name' during bokeh range computation. values and range fail the same way with the same probe.

This adds a _resolve_dimension helper that falls back to looking the dimension up by name — the string lookup path already matches both names and labels via its name_map — and uses it in isscalar, range and values. Also drops a leftover duplicate get_dimension call in isscalar.

The regression test covers the three interface methods plus the render path from the issue.

gdf = gpd.GeoDataFrame({'status': ['safe', 'endangered', 'safe'],
                        'geometry': [Point(0,0), Point(1,1), Point(2,2)]}, crs='EPSG:4326')
points = gv.Points(gdf, vdims=[('status', 'Conservation Status')])
points.opts(color=dim('status').categorize({'safe': 'green', 'endangered': 'red'}))
hv.render(points)  # AttributeError before, renders after