The study actually does show small positive effects to alcohol, but it's also trying to differentiate alcohol (minimal impact at low doses) from smoking (strong and strictly negative impact at all doses). You can see the study's alcohol results by dose here:
https://www.bmj.com/content/bmj/368/bmj.l6669/F2.large.jpgFrom the study text:
> We observed a relatively small difference in life expectancies across different levels of alcohol consumption compared with other individual lifestyle factors. The cardiovascular benefits of moderate alcohol consumption have been consistently observed in large cohort studies, but alcohol consumption and risk of cancer showed a dose-response relation. Thus, current guidelines do not encourage a non-alcohol drinker to start drinking just for the benefit of preventing cardiovascular disease.
Female life expectancy peaks in the two lowest non-zero ranges, and disease-free expectancy at the second-lowest specifically; male life expectancy rises all the way to the top of "moderate" consumption, and disease-free peaks at the top two "moderate" doses. Presumably that's partly from men being larger and metabolizing alcohol faster, and partly from men being at higher risk of heart disease.
(There's probably a healthy-user bias here too, since there are conditions and medications which weren't checked for that would make people unable to tolerate alcohol, but that's less relevant when comparing different doses.)