Consider switching endemicity to state encoding

[TheoPannetier]

As of now, following the Tutorial and the MuSSE example, we have the following encoding:

• not_present = 1
• nonendemic = 2
• endemic = 3

This is arbitrary, works well. I have hard-encoded the convention into all_endemicity_status().

However, GeoSSE requires a different encoding:

• Area A only (mainland, not_present) = 1
• Area B only (island, endemic) = 2
• Area AB (widespread state,nonendemic) = 3

The output of BioGeoBEARS also appears to follow this convention. I'll switch the current convention to this one to avoid the coexistence of two encodings and the risk of confusion.

[TheoPannetier]

Done, values switched!

[TheoPannetier]

Correction: GeoSSE requires states 0, 1, 2, with 0 the widespread state. MuSSE requires states 1, 2, 3 (any can be the widespread, since it's considered as just a different state). BioGeoBEARS follows the convention 1, 2, 3 with 3 the widespread state in its output.

Because of this inconsistency I can't make a single switch for all models. I'll add an argument model to endemicity_to_states to return the desired output for MuSSE / GeoSSE

[TheoPannetier]

Done