Well, your insurance policy works by seeding humans across great gulfs of time and distance, insulating them from single-event catastrophes. It seems the parameter to optimize is thus the product or sum of space and time. By that measure, getting to Alpha Centauri faster might not be an actual improvement.
But if we switch to looking at two states in time, humanity at t0 and humanity at t(year), we'd want to maximize the distance between clusters of humans and the total number of seperate clusters. In that case, if we had an improvement in travel velocity, we wouldn't use it to reduce the time, we'd use it to increase the distance.
We'd also not send multiple ships to the same destination, which wouldn't add to the number of seperate clusters, we'd scatter them out to cover as many destination stars as possible.
Since the greatest threat to their long-term survival comes from humanity itself (or its remnants or replacments), it also indicates that many of these ships should go long, go deep, change course multiple times, and run silent, just like they were hiding from an alien species bent on human extermination.