Data Model doesnt consider the subcategories of Pokemon

As briefly mentioned by me in #1026 , the data model doesnt consider the different subcategories of Pokemon. While up untill Sun and Moon, the disction "is_legendary" was possible, it always disregarded the mythical moniker, and in sun and moon the moniker sub_legendary was introduced in code, into which ultra beasts also were put into.

And Scarlet and Violet/Gen 9 now further split this into Ultra Beasts, Sub Legendary, Paradox Pokemon, Restricted Legendary and Mythical.

Therefore the datapoints of is_legendary is much more detailed than is previously assumed. One could also argue lorewise while Necrozma is a Restricted Legendary, its also is a Ultra Beast and much more. Koraidon and Miraidon are called paradox pokemon, but are handled as restricted legendaries in code.

So, instead of strict boolean is_legendary and is_starter, a list of categories a pokemon is in would be more appropiate.

The issue names no files, tests, or entry points; begin by locating the current data-model definitions for is_legendary and is_starter. Done means the representation can express the listed Pokémon categories, but the issue does not specify an implementation target or validation test.