P and Q indicate the *type* of the entity ("P" = "Property", "Q" = "Item" for
arcane reasons), "L" = Lexeme, "F" = Form, "S" = Sense, "M" = MediaInfo). As you
can tell, we'd quickly run out of letters and cause confusion if this became
configurable.
I don't think this would cause a confusion, because the lexicographical project is really a separate project that just happens to reside on the same Wikidata domain. Essentially you did internally what we are asking for other sites - you mixed two projects, and kept them distinct by using different prefixes. If at some point you decide to add some new area of data, e.g. biological, you could add new prefixes for that too, but that would also be a "separate" project.
Most other sites that link to Wikidata only care about just one of those projects. E.g. OSM would have very little interest in lexical data, so it is OK if "L" prefix would be used in OSM and in WD because it won't be as confusing to the users as reusing the Q.
The prefix can be omitted for local entities, so Q12345
is an item on the local repo (or the default repo of a wikibase client).
I think that was a big mistake -- the "(or the default repo of a wikibase client)" -- because wd implies Wikidata, not Wikibase, so it dilutes the meaning of "wd:". See my other email on how I fixed it.