Fix entry mapping

[liach]

For long, Entry mapping can be null, indicating an absence of mapping. Entry mapping always require a mapped name to be present; however, besides mapped names, it carries other information, mainly access modifier change and javadoc.

As a result, you cannot have standalone javadoc information without mapping. This has become problematic, and a solution is to revamp all entry tree usages from nullable entry mapping to non-null no-op entry mapping instead, and we can revamp entry mapping itself to carry javadoc information without a rename.

[Runemoro]

I think #188 fixes this?

[modmuss50]

Looks like it does.

It will need testing.

[liach]

That pr misses a lot of places, as in existing enigma code there is widespread assumption that target name cannot be null in an entry mapping, and that pr apparently takes care of none of that. Also that pr allows an entry mapping to be effectively same as null by having null javadoc + null target name, yet that pr does not handle that at all (still using null checks without checking members)

[YanisBft]

Fixed by #364 it seems