The linked article is quite thought-provoking. However, the claim is not quite clear: What is shown in figure 3 is a short segment of the genome (less than 6%). Over this region, Figure 3 shows much less apparent evolutionary selective evolutionary pressure in a comparison of one pair of selected viral sequences (panel B) than in a comparison of another selected pair (panel A).
To elaborate: For the selected pair in panel B, there has been the same amount of apparent neutral evolutionary drift and a much lower amount of apparent selective pressure over the region discussed.
This can be interpreted as:
(A) The noted region is restrained from mutating in the second pair specifically, for example if it has naturally acquired a new function that gives the virus a new advantage, and this new function requires the sequence in unchanged form.
(B) The noted region has had less pressure to mutate in the second pair specifically. For example, there has not been the same difference in evolutionary :hardship" as the first pair. If you take two pairs of twins and the first pair were raised in the same household, went on to similar life experiences, they may end up looking similar. If, for the second set of twins, one twin went down t'pit and the second one married a lord, they would end up looking different. So maybe the pair of genomes that look "too similar" in mutational profile in the second set, had simply co-existed under circumstances where this region of the genome was not forced to adapt? The amount of data shown is too small and selective to be able to tell.
(C) Could there have been genomic "recombination" in the second pair, which is an entirely natural phenomenon. This is where genome segments mix-n-match between different viruses with different histories.
(D) There has been codon-optimization in a lab, but if so, it should be pretty obvious, from the sequences, what kind of codon optimization was done. It is also not clear why it would be done - there is no need for it, and yet it leaves a fingerprint of intervention.
I don't know why this person's analysis was not done comprehensively on all regions of all genome pairs, to show which members of which pairs are really the outliers. Right now, either member of either pair could be the outlier, and it seems counter-intuitive to make a claim based on this, when it is so easy to repeat this analysis on other pairs of coronavirus genomes to isolate the outlier and thereby strengthen the evidence. All the data are readily available - in fact for thousands of COVID genomes now.