HMTK mapping tool and tutorial update

[kejohnso]

@mmpagani

This issue describes two current problems in the HMTK, its manual/tutorial (this repository), and training notebooks built on the HMTK, as well as our plan to remedy these problems.

We are creating updated training material (a manual for building PSHA models + notebook-based examples) as part of the TREQ project. This work will include updating the notebooks used in workshops to demo the HMTK, and raises the following issues:

1. The source model parser used by the notebooks is now deprecated and points to the hazardlib to_python .
2. HMTKBaseMap was used by the workshop notebooks and relied on Python's basemap, which will be deprecated, and thus HMTKBaseMap was removed in OQ Engine v3.9; thus we need new tools for mapping.

to_python converts the sources to hazardlib classes (simple fault example), whereas the HMTK tutorial and accompanying training notebooks use mtk classes (simple fault example). Additionally, the hazardlib classes include more source typologies than the mtk classes. In order to ensure full consistency between the HMTK and the hazardlib, all new and updated training material will use only the classes of the hazardlib, which are also used by the OQ Engine. This change will require a major update to the HMTK Tutorial (this repo), and so we will produce a version 2.0. The version 2.0 updates will prioritise issues raised here, but will also try to include other necessary modifications flagged by users over the past few years.

We must create an updated version of HMTKBaseMap and its plotting methods. After much discussion, the hazard team has decided not to replace mapping.py, but instead to create mapping tools in the OQ-MBTK that are designed to work well with the HMTK. The primary reason for moving these plotters to the HMTK is that the plotters will use GMT 6.0, and we do not want to make GMT an OQ Engine dependency. This change applies only to geographic plotters; all other HMTK plotters that depend on matplotlib will be kept.