design of SERIALIZED_LWGEOM/LWGEOM/PG_LWGEOM

I am reading postgis code of function LWGEOM_in:
in older version: wkt -> PG_LWGEOM
in newer version: wkt -> SERIALIZED_LWGEOM -> LWGEOM -> PG_LWGEOM

So my question are:

(1)SERIALIZED_LWGEOM and PG_LWGEOM have the same serialized format except for the member of size's position?
(2) if SERIALIZED_LWGEOM for code isolation of parser and postgis, why not use
wkt -> SERIALIZED_LWGEOM -> PG_LWGEOM directly?

any idea?