Pylint reports the following warning.

pygmt/datasets/samples.py:351:0: W0613: Unused argument 'kwargs' (unused-argument)

I think we don't need kwagrs for new datasets.

I'll check again to confirm after I rewrite the tests, but I think it needs **kwargs because of data = load_func[name](suppress_warning=True). It was getting passed a value for suppress_warning but wasn't expecting any arguments.

I received TypeError: _load_earth_relief_holes() got an unexpected keyword argument 'suppress_warning' when I tried running it without **kwargs.

The suppress_warning parameter (or **kwargs) was added so that users who call the deprecated, old load_xxx functions can silent the deprecation warnings by passing suppress_warning=True. Thus, **kwargs is NOT required for new private _load_xxx functions.

Do you think we should rewrite the code to call the old and new load_xxx functions in different ways:

    # old public load functions for backward compatibility
    load_func_old = {
        "bathymetry": load_sample_bathymetry,
        "fractures": load_fractures_compilation,
        "hotspots": load_hotspots,
        "japan_quakes": load_japan_quakes,
        "mars_shape": load_mars_shape,
        "ocean_ridge_points": load_ocean_ridge_points,
        "usgs_quakes": load_usgs_quakes,
    }
    # new private load functions
    load_func = {
        "earth_relief_holes": _load_earth_relief_holes,
    }
    if name in load_func_old:
        data = load_func[name](suppress_warning=True)
    else:
        data = load_func[name]
    return data

I think that's a good idea. I think this should be a separate PR; does that work? I'm happy to do it (admittedly copying the code above) but won't be able to get to it until tomorrow.

A separate PR is OK to me.