source) shows quite nicely that according to current understanding, Africans are nested within Eurasian genetic variation. The modern humans have the following structure:
(Early modern human lineage detected as admixture in the Altai Neandertal, ((Asians, Europeans), Africans)),
and then there are two deeper layers of Eurasian hominins (Neandertal/Denisovans) and the "Mystery hominin" that mixed into Denisovans.
Africans are thus just a leaf of the Eurasian family tree, casting serious doubt -if this model is to be believed- to the position that H. sapiens originated in Africa and are descended from people who never left the continent. It seems much simpler to derive them from an early migration (~200kya?) from Asia which would nicely explain why the continent's first sapiens populations appear tentatively in the northeastern corner, and why they do not replace archaic hominins for most of the 200 thousand years until today. In a reversal of perspective it is not Skhul/Qafzeh that are the "migration that failed", but rather the Omo 1 outlier is.
One might argue that this is just a consequence of the fact that lots of ancient genomes have been published from Eurasia, but none from Africa. So, there are all these branches of deep archaic Eurasians simply because there are no genomes of deep archaic Africans.
But, this explanation does not really work. If Africans had any significant ancestry deeper than the split of "Early modern human lineage", then this lineage would be closer to (Asians, Europeans) than to Africans. However, Kulhwilm et al. assert that it is "equally related to present-day Africans and non-Africans". If they had any ancestry deeper than ((Denisovans, Neandertals), H. sapiens), then (Denisovans, Neandertals) would be closer to non-Africans than to Africans. Well, they are, but this is now satisfactorily explained by admixture from (Denisovans, Neandertals) into non-Africans, thanks to genomes like Ust Ishim, K14, and Oase which have big chunks of Neandertal ancestry that can't be explained any other way. No need to invoke any such lineage when a simpler well-documented alternative exists.
The presented phylogeny negates the possibility of the existence of collateral archaic African kin of the extant Africans that admixed with them, and leads to the conclusion that Africans are nested within Eurasian variation because they really are. This is, of course, incompatible with the statistically inferred archaic introgression into Africans which indeed postulates the existence of such archaic Africans and their contribution to extant ones.
I don't see any obvious flaw with Kulhwilm et al. but if its model is right, then it does lead to some rather extreme conclusions. It contradicts the evidence for archaic introgression; if Hsieh et al. is wrong (and I don't seen any evidence for that either), then Kulhwilm et al. can be saved, but only if Africans are really nested within several layers of Eurasian variation and did not admix at all with the morphologically diverse archaic Africans of the paleoanthropological record. This also doesn't seem right now that we know that sapiens-archaic admixture was a common occurrence in Eurasia. The reversal of perspective alluded to above may help here by removing the opportunity for admixture, but that too is, of course, an extraordinary claim.
In sum, I am rather convinced that the latest discoveries have muddled the origin story of our species and some major rethink is needed to evaluate the totality of the evidence.