One could probably compile remarks like that and get a pretty good image of what an aristocratic Roman considered a normal lifespan for an aristocrat
Yes, although these sorts of remarks can be difficult to generalise from. In the first place, life expectancy from birth is not an especially meaningful metric for any particular person's experience of age and death. It can vary significantly within a population (there is a 7 year disparity between the highest and lowest life expectancy by state today in the USA, 12 years if you go highest female to lowest male) and when its being so massively offset by child-mortality, it won't reflect the way that people perceive aging in society, because a life expectancy in the 20s won't mean that people are typically actually dying in their 20s. This is all before we get into the vagueries of interpretation. For example, while 63 would probably not be viewed as "not untimely", but certainly we don't need to get to 79-80 for many people today to have such a reaction. (I'd imagine that we could find opinions along those lines being voiced about some as early as their late 60s and certainly by their mid-70s.)
There are a number of medieval comments like this as well that are equally interesting, though no less problematic to interpret.One of the more famous is William of Malmesbury's comment about his age in the introduction to his commentary on Lamentations: "Quadragenarius sum hodie, admouique pedem <ad> medietatem metae quam diuinus psalmista ponit hominum uitae, dicens: Dies annorum nostrorum in ipsis septuaginta annis. Si autem in potentatibus octoginta anni, et amplius eorum labor et dolor. (Today I am forty, and I draw near the midpoint of the course that the divine psalmist set out for the life of man, saying: The days of our years in them are threescore and ten years. But if in the strong they be fourscore years: and what is more of them is labour and sorrow. [Ps. 89.10])
an area (Fayyum) which 'may have been unhealthy even by the low standards of the region.' ... the findings, as the paper acknowledges, can only shed light on a specific place and time
Oh no doubt! And indeed he finds a life expectancy of 22 for Women (who are apparently by far the best represented in the date) in that region. But it's not like the data we have for Rome is radically different. Both information about membership of the Senate and certain city councils as well as Ulpian's life table suggest that life-expectancy in Italy was likely also in the 20s, although Scheidel seems to prefer an estimate in the mid- to high-20s or just around 30. (This would not be out of keeping with the modern comparison I note above, where regional variation likewise ranges between 5-10 years, although the variation within Europe which is closer to 15 years (or 13 if we restrict ourselves to the EU) is maybe a better comparison.)