Updated March 15, 2019 and March 20, 2020
with a new interpretation of the skull sutures of the stem tetrapod Tiktaalik. Now, with the addition of more closely related taxa, Koilops is a finned fish that nests basal to Elpistostege, Spathicephalus and Tiktaalik off the main lineage of the fin to finger transition.
Figure 2. Koilops is a flat-headed sister to Spathicephalus, but with teeth, larger orbits and a shorter snout
Koilops herma (Clack et al. 2016; NMS G. 2013.39/14) Tournasian, early Carboniferous ~375 mya) is a temnospondyl with a flat skull and large orbits nesting between Greererpeton and Spathicephalus. The nares were close to the rim of the short rorstrum. The pineal foramen was enormous. The teeth were small and sharp. The nasals were broad.
References
Clack et al. (14 other authors) 2016. Phylogenetic and environmental context of a Tournaisian tetrapod fauna. Nature ecology & evolution 1(0002):1-11.
The Pterosaur Heresies 
Why do you give laterally-facing eyes to this extremely flat bottom-dweller? And why do you call it a filter-feeder when there’s no evidence for a filter apparatus? You can’t strain anything with this kind of dentition.
It’s more than that, or its position as a baphetoid wouldn’t be so well supported in my preprint; and there’s no such thing as a “main” feature.
(Also, baphetoids aren’t temnospondyls at all.)
Eucritta very strongly suggests that no fusion ever happened: instead, the orbits gradually developed an extension.
Laterally facing eyes are standard tetrapod features. Even with flat headed taxa. Like flounders. I think you know that. You might have forgotten or have been swayed by star-gazing restorations. Perhaps that’s why you read the PterosaurHeresies, to ‘think different’ (ly).
When I say ‘temnospondyls’ that’s because baphetids nest in the LRT at the base of the temnospondyls. Did I overextend the clade?
re: Antorbital fenestrae: I agree, enlargement could have happened. Here ‘fusion’ is short hand for two holes merging regardless of their origin.