I would say it is directionless. Neither forward nor reverse. The fluctuations in life expectancy can have many other factors other than ageing itself being under the umbrella of selection.
Moreover, life expectancy itself is not a great readout for ageing.. The ageing itself might have occurred at the same rate as it does now. For most part life expectancy measurements did not result from the fact that people died because of old age at 25 or 35. But infant mortality rate was much higher. Think about it Aristotle died at 62 and Socrates at 70 (due to infection and murder respectively, rather than age associated disease). They were not ageing more or less than us, even though modern medicine increased our life expectancy. Hence, I think using life expectancy fluctuation over different eras of human evolution, doesnt say much about the process of ageing.
As far as ageing itself is concerned, goal would be to maximise survival, but that doesn't come without trade-offs. What I am trying to say is that it is likely that the reduction of ageing process(if observed) is not goal, but result of environment constraints.
For instance you mentioned that ancestors had a stronger immune system. Highly likely. Now as per one hypothesis the stronger inflammatory responses to pathogens might be more a bit towards faster ageing. Now as per some studies (though I still need to see more convincing paper), the African ancestry may age faster. Nonetheless, a sub optimal immune response in Africa might have decreased your life expectancy anyway.
What do you suggest?
What we are looking at is the average life expectancy, not just individual ages.
Moreso, we considered the other hominid ancestors of the extant Homo sapiens - like the Australopithecus; not just the ages of people in this era.
These average ages are; better looked upon; as something that spanned over many millennia.
Like the period of the earlier hominid specied to the Neolithic era isn't just some few years. If it is within a short period of time; then it might not necessarily be termed evolution - though some theories support catastrophism instead of gradualism. But overly; evolution is considered a slow process, which spans millions of years.
I know we are talking about average life expectancy. However, this variable is very different from ageing. Ageing effects average life expectancy of population under measurement. Let's say you are talking about average life expectancy at age 35 or 70. What it will end up telling you is a chance that most of the people in this population will live upto that age. See this data for instance. You can take into account the life expectancy at different ages. McDonald and Ruhe explains this difference very nicely, in their review publication in journal nutrition. They also talk about the evolutionary forces acting on ageing.
In fact we can ask the question that how much of average life expectancy is under the regulation of genes? In fact twin studies show that genetic contribution over here is just some 20-25%. While rest of the variation is explained by environment. I think that 20-25% should be contribution of ageing. But what is under selection? Is slow ageing selected against? Or some other benefit that also accelerates ageing tends to increase fitness? For instance take p53 selected which maintains genomic stability. However p53 achieves so by inducing either cell cycle arrest or apoptosis. Both of these can affect the ageing phenotype..
For the second part, as long as you happen to select a trait, it will fall under umbrella of evolution. And evolution can happen even as a step function. There has been huge bottleneck events in past that have shaped many species. Selection of lactose tolerance trait and fair skin in European populations did not take millions of years either. I think it took something like 8000 years.