Ceratosaurian ornithomimosaurs - another crazy idea?

The first taxon to be added to Ceratosauria after Ceratosaurus itself was Ornithomimus (Marsh, 1895).  While this version of Ceratosauria has been long forgotten since Gauthier resurrected the name in 1986 for Ceratosaurus and coelophysoids, is it possible Marsh was on to something?  While recoding Holtz's (1994) matrix, I started noticing the coelurosaur-like characters of Elaphrosaurus.  For instance, the anterior cervical centra have very low anterior articular surfaces, there is no post-obturator notch on the ischium, the pubis lacks an obturator foramen, and the fourth trochanter is reduced.  Of course there are loads of more basal characters too, which is why it still clades in Ceratosauria in good analyses like Rauhut's (2003).  But what if we took another approach and placed ornithomimosaurs next to it in Ceratosauria instead?  Crazy, you say?  Probably, but it could help explain some otherwise incongruous features of ornithomimosaurs-

- Basal taxa like Pelecanimimus and Shenzhousaurus have very long antorbital fossae, and Shenzhousaurus has a longer portion of it taken up by the antorbital fenestra than most other coelurosaurs.
- Ornithomimosaurs' supposed maxillary fenestrae are usually small, not necessarily invasive, and resemble the sub-fossae of taxa like Ceratosaurus and Carnotaurus in being dorsal/ventral to their promaxillary fenestra.  Senter (2007) doubted their homology to maxillary fenestrae.
- Pelecanimimus is unlike coelurosaurs closer to birds than tyrannosaurids in having lacrimal horns.
- Ornithomimosaurs are like Limusaurus and coelophysoids in having reduced axial parapophyses and diapophyses.
- The perhaps basalmost known ornithomimosaur Deinocheirus has a short posteroventral coracoid process, unlike other maniraptoriforms.
- The humeri are straight with reduced deltopectoral crests, of course.
- The small carpals without obvious articular surfaces make more sense in a ceratosaur than a coelurosaur.
- Deinocheirus has a manual phalanx III-3 shorter than III-1 + III-2.
- The brevis fossa is broad and triangular unlike most coelurosaurs, but like non-tetanurines.
- The ilium's posterior edge is notched as in ceratosaurs.
- The ilium has a peg-in-socket articulation with the ischium in both groups.
- The pubis is directed further foreward than in most coelurosaurs.
- Shenzhousaurus seems to have an obturator foramen in its ischium, as in Limusaurus.
- The ischial boot is large as in ceratosaurs, but unlike any coelurosaur.

That's all great, but don't ornithomimosaurs have a ton of tetanurine, avetheropod, coelurosaur and maniraptoriform characters?  Well sure they have some, like fourth manual digit absent and metatarsal III proximally reduced.  But it's not often realized just how many features usually thought to be exclusive to coelurosaurs are also/already found in more basal taxa.  Ceratosaurus has pneumatized paroccipital processes.  Majungasaurus has a palatine-pterygoid subsidiary fenestra.  Ceratosaurs have distal caudal prezygapophyses over half of central length.  Majungasaurus has only eighteen caudals with transverse processes, similar to Tyrannosaurus and Compsognathus.  Carnotaurus has ossified sternal plates, which are otherwise only known in Baryonyx among non-maniraptoriforms.  Many ceratosaurs have tall astragalar ascending processes, and they've long been recognized as having large pubic boots and aliform anterior trochanters.

Unfortunately, a published matrix to test these ideas in doesn't yet exist.  While Xu et al.'s (2009) matrix (based largely on Smith et al.'s) would seem to be a good candidate since it includes Elaphrosaurus, Limusaurus, Ceratosaurus, Majungasaurus, Shenzhousaurus and an ornithomimid (though lacking Pelecanimimus and Deinocheirus), their habit of not coding taxa for "irrelevent" characters means they'll never find support for heterodox topologies such as this.  Majungasaurus is miscoded as lacking the subsidiary fenestra, Shenzhousaurus is miscoded as having a small antorbital fenestra within the fossa, ornithomimosaurs' possible maxillary fenestrae are assumed to be homologous to tetanurines', ornithomimids are coded as having large axial parapophyses and not coded at all for axial diapophyses, ornithomimosaurs aren't coded for distal carpal size or fusion, the ornithomimid wasn't coded for the postacetabular notch, Shenzhousaurus was miscoded as having an obturator process, the ornithomimid was miscoded as having a post-obturator notch, both ornithomimosaurs were miscoded as lacking an ischial boot, and both Elaphrosaurus and ornithomimosaurs were miscoded as having stout, well-developed fourth trochanters.