Adds deprecated as a node property slot added to biolink:Entity

[kevinschaper]

I was surprised to see that the model doesn't yet have a way to represent that an entity is obsolete or deprecated. I could definitely use a double check on my exact_mapping to oboInOwl:ObsoleteClass and I'm not sure if we want to make the slot more targeted than allowing it for any named thing.

[RichardBruskiewich]

Another option would be to place the slot on the (abstract) biolink:Entity class which covers both biolink:NamedThing and biolink:Association classes.

I don't know how often retention of a piece of knowledge when deprecated would apply to an association instance - a knowledge statement - per say, except that perhaps it is a way of flagging that a knowledge statement, recently previously considered "true", may now be deemed "false". Of course, one can issue a second statement saying this; however, perhaps there are circumstances when the evolution of knowledge can be more accurately captured in this fashion (much as the evolution of ontology is)

[RichardBruskiewich]

I recommend that the 'deprecated' slot be moved upwards to biolink:Entity. This also requires removal of the is_a: node_property as well. That may make it a bit more elusive to detect in the code but I think we'll be able to (quickly) wrap this within the various Biolink libraries (BM toolkit, KGX, etc.)

As discussed over Slack, the other thought that comes to mind is to also add a "replaced by" slot to allow the deprecated entity (instance) to point to its replacement.

Admittedly, this latter idea is more feasible with entities with stable identifiers (as found in reference databases, e.g. gene loci) and harder to envision for knowledge statements (which don't have stable identifiers) but perhaps with some flexibility in the range of value and perhaps, with the use of clarifying slot_usage description comments, it may still be made more broadly useful.

The 'replaced by' functionality may already be managed in some contexts by entity identifier versions (i.e. the Genbank accession identifiers version suffixes come to mind...). I don't know how fine grained we want to get here in representing such information in KG's but it's a thought.

The range data type for 'replaced by' could just be 'string' for now (it is tempting to use uriorcurie but that may be too restrictive?)