Use the define rather than a raw number

Arguably this isn't strictly accurate, since libnmg I/O versioning isn't
TECHNICALLY tied in a data sense to .g I/O versions, but functionally it should
be fine and provides at least a little context.

The broader problem here is needing to use versioned librt routines in an API
that only really wants to be working with rt_db_internal structs.  Wondering if
it wouldn't be cleaner to encode versioning of the idb_ptr data in the
rt_db_internal container rather than needing to propagate the dbip for that...

include/rt/db_internal.h

@@ -42,6 +42,11 @@ struct rt_functab; /* forward declaration */
 
 /**
  * A handle on the internal format of a BRL-CAD database object.
+ *
+ * TODO - right now, rt_db_internal doesn't encode any notion of data version,
+ * which appears to be a major reason some APIs need to pass a dbip through in
+ * addition to the rt_db_internal.  Should we add an idb_version entry here to
+ * indicate idb_ptr data versioning?
  */
 struct rt_db_internal {
     uint32_t            idb_magic;

src/librt/primitives/nmg/nmg.c

@@ -1633,7 +1633,7 @@ rt_nmg_mat(struct rt_db_internal *rop, const mat_t mat, const struct rt_db_inter
 	return BRLCAD_ERROR;
 
     // Import 
-    struct model *nm = nmg_import(&ext, mat, 5);
+    struct model *nm = nmg_import(&ext, mat, BRLCAD_DB_FORMAT_LATEST);
     nmg_km(m);
     rop->idb_ptr = nm;