Four tilemap tests fail likely due to upstream changes

[seisman]

See the failures in the recent CI runs: https://github.com/GenericMappingTools/pygmt/actions/runs/6270258961/job/17027947064

Here is one failing test:

Expected	Actual

Likely some changes in the original tiff files returned from the tile server?

[weiji14]

Likely some changes in the original tiff files returned from the tile server?

Ah yes, Stamen has moved their tiles to Stadia and that came with a bit of a redesign, see https://stamen.com/here-comes-the-future-of-stamen-maps. We should also see what contextily is doing - geopandas/contextily#220, or switch to a non-Stamen basemap (that is somewhat stable) for the unit tests.

[weiji14]

We'll also need to change the second example at https://www.pygmt.org/v0.10.0/gallery/maps/tilemaps.html that uses Stamen Watercolor, after Stamen stops redirecting to Stadia's servers on 31 October 2023 according to https://maps.stamen.com/stadia-partnership.