These concerns are addressed at least to some extent in the paper.
Note also, the title not necessarily means "resistance impossible", it may mean "currently no resistance, even after explicit attempts to evoke it, using methods that for other antibiotics do result in resistance".
I myself am not competent to judge. You may be right in your skepticism.
From the paper:
> The evolution of bacterial resistance to COE2-2hexyl was analysed in S. Typhimurium and MRSA, using either serial dilution or morbidostat-based experimental evolution (see Methods). Serial dilution. No high-level COER mutants (≥50 × MIC) of either S. Typhimurium 14028 or CA-MRSA USA300 were isolated after serial dilution for 21 days; however, low-level COER mutants were recovered in both organisms (8 × MIC and 4 × MIC, respectively) (Supplementary Fig S1a; Supplementary Table S4a and b). Morbidostat. Although high-level CIPR mutants of wild-type E. coli (BW25113) were obtained after morbidostat culture for >3 days (64-128 × MIC),35 no COER mutants were observed under these conditions. Therefore, E. coli mutL (JW4128) and wild-type A. baumannii ATCC 17978 were assessed for COER mutants since both strains possess an inherently higher frequency of mutation (∼10-100-fold), resulting from deficiencies in either methyl-directed mismatch repair or the intrinsic DNA damage-inducible response, respectively.45,46 Only low-level COER mutants (2-4 × MIC) were recovered from either of these hypermutable strains, which is in marked contrast with the high-level CIPR mutants (128 × MIC) observed for A. baumannii35 (Supplementary Fig. S1b; Supplementary Table S4c–e). Whole genome sequencing of COER mutants revealed mutation(s) in genes encoding essential functions involved in membrane remodelling and secretion (Supplementary Table S4a–e). Examples include: Gram-negative secA and bamA involved in outer membrane protein biogenesis and lptD from the lipopolysaccharide translocation complex47,48; pgsA from the phosphatidylglycerol biogenesis pathway49; and Gram-positive pmtR from the cytolytic peptide toxin transporter pathway.50 Consistent with this observation, a de novo constructed S. Typhimurium secA nonsense mutant (G880∗ (GGA→TGA))—derived from COER mutant analysis from serial-dilution—showed a two-fold increase in COE resistance (Supplementary Table S4a). Taken together, these data suggest that COE2-2hexyl did not evoke significant bacterial resistance, potentially due to specific effects on essential functions involved in membrane remodelling and secretion.
[... they go on in the Discussion section ...]
BTW, the "serial dilution" method name sounds to me like what is illustrated by this famous video: https://www.youtube.com/watch?v=plVk4NVIUh8