I noted that it simply defines "light drinker" as <= 1 drink per day. It's not clear to me from reading the study how someone who drinks once every two or three days, or once a week, would be classified. The authors note some of the studies they incorporated in their results considered occasional drinking as "non-drinker" while others did not, so it seems that it was not consistent. However, it seems that in some cases the "non-drinker" category means people who completely abstain from alcohol.

Presumably you're right, they would be less likely to smoke, and I would imagine also possibly more health conscious in other ways. The authors mention that a different analysis of 15 studies of alcohol use among never-smokers was similar to their own findings, but also caution that tobacco could indeed be having an influence on their results (or even a possible synergistic effect with alcohol).