In short: none that I could find. Indeed, it's a very one-way relationship - the more you drink the worse reaction time gets. However, there is a starting point, which is generally put around .02-.06 depending on your chosen study - below that threshold the effect of BAL on reaction times is difficult to pin down. Most studies I found showed that below .05% often showed no change in cognitive functioning regarding reaction times, though it was often theorized that intoxicated people are able to compensate for their impairment under normal testing conditions, as in this study: Alcohol increases reaction time and errors during decision making
This suggests a possible answer, however - you knew if intoxicated your reaction times could be worsened, so you focused extra resources on maximizing them.
Also, studies in general are statistical in nature - none say that "no one ever showed improved reaction time while intoxicated", because sometimes its just a fluke, sometimes people are especially trained at compensating for intoxication (even at a subconscious level), and a host of other similar factors.
What this all indicates is that even if your reactions were repeatedly tested under measured circumstances, you might very well be able to show improved reaction times under varying level of intoxication; however this is extremely likely to require more of your mental resources, so that under stress or complicated situations you would find real-world performance to be poorer.
Finally, I would also warn that reaction time does not necessarily correlate with any/all other cognitive tasks. Those with even .05 BAL are often found be twice as likely to be in an automotive accident, even though most reaction time studies and general cognitive ability tests would show low/no impairment.
But, still I cannot say there is any literature I can find that studies specifically the effect of low-moderate levels of intoxication on a variety of individuals, especially with the goal of identifying potential enhancement of any cohort within the study. At least if there was such a study it could have been considered to have failed to find any result worthy of publishing, or be rephrased to be a study that showed reduce reaction time in fitting with other similar research.