Expand documentation on graph projections and allow finer grained control

Nobody has claimed this yet.

Starting with #806 we have better documentation of OWL graph projections, but this could be improved, and aligned with other systems. There are also some potentially confusing aspects in that RG uses a reduced projection profile, so entailment results can look different from direct.

The challenge here is that as GPs become more inclusive, the more confusing a single triple becomes - longer term the solution here is to use a PG standard like owlstar to annotate edges.

current docs:

https://incatools.github.io/ontology-access-kit/guide/relationships-and-graphs.html#further-notes-on-owl-and-graph-projection

OWL Axiom | Graph Projection | SQL Adapter | Relation Graph
-- | -- | -- | --
A SubClassOf B | rdfs:subClassOf B | Yes | Yes
A SubClassOf R some B | A R B | Yes | Yes
A SubClassOf R value B | A R B | Yes | No
A SubClassOf R only B | A R B | No | No
A SubClassOf R max 0 B | A R B | No | No
I type A | I rdf:type A | Yes | No
I type R some A | I R A | Yes | No
I type R value A | I R A | Yes | No
I Facts: R J | I R J | Yes | No

This doesn't include OBask neo4j2owl https://github.com/OBASKTools/neo4j2owl/blob/master/README.md#owl-2-el---neo4j-mapping-direct-existentials cc @hkir-dev

Start with the linked relationships-and-graphs documentation and review the existing OWL graph projection table, including the reduced Relation Graph profile and differences from direct entailment. Compare the listed systems with the linked neo4j2owl mapping; done means the documentation explains these projection differences and the requested finer-grained control clearly.

Written by the indexing model from the issue text.