Neighbourhood visualisation/optimisation

[wlason]

Hi! Thanks for developing this cool package!

I was just wondering if you had recommendations for n_neighbors and radius parameters passed to compute_spatial_niches()? It is difficult to assess if the neighbourhood graph makes sense if it can't be seen. Can I use sq.pl.spatial_scatter()?

I found trimming the longest links in the neighbourhood graph helps for my dataset, can I use precalculated neighbours in quiche?