entry_refs inside BMessages aren't quite handled correctly

[anevilyak]

Making a ticket so this isn't forgotten. Currently, when the BMessageValueNode encounters an entry_ref field, it maps that to be handled by the corresponding DWARF type for the actual entry_ref structure. However, this isn't entirely correct since the flat format differs slightly from the in-memory format (the latter has the name field as a pointer to an allocated string, whereas the former packs an array directly into the structure). Consequently the name field isn't displayed properly. This needs to be fixed by adding debug_info support for creating/registering synthetic types.

[luroh]

Moving Debugger related tickets out of R1 milestone (Prop #14).