An interval and an imprecisely known instant are not the same thing. A person with a lifespan of 1903/4 through 2016-03-17 lived during a period of time where one end of the interval is not fully known (so is, itself, an interval), so the duration of the interval is not known. The interval representing the bounds of imprecision for the birth date, and thus the lifespan interval, serves a very different purpose in my mind.
I don't think the concept of "lifespan" as an event makes sense. It's a closed interval with two terminal events, even if the dates of the events aren't known with precision. There are related events such as "practicing in the mid-1930s" or "dead by 1927 when his will was probated" or "alive in 1637 when she gave testimony in court," but these events just help provide some maximal/minimal bounds for the terminal events in the lifespan.
Tom