Waiting for replication of the study is practically a necessity before trying to make any sweeping policy decisions.
At the same time, a cautious person may decide that there's a probability that this correlation is both true and also a causal association. And in caution, stay below the one drink per day cutoff in the study, or even go to zero, when pregnant. Already, there are many guidelines on for for pregnant women that are based on abundance of caution rather than full knowledge of definite risk.
TL;DR the scientific literature is not a textbook, it is the active working document and communication between scientists as they are working through the self-correcting scientific process. And the scientific literature should treat any single paper as a collection of data that is incomplete, may not be fully understood, and may have faulty conclusions attached to the data. It takes time and further evaluation do fully understand natural phenomenon. The rare paper that conclusively proves something is the exception, and in the end it is only proven when it has been replicated by others.