In Gustavus Fischer's Details of Syntax, a number of pages are devoted to the uses of tenses in temporal clauses. It is fairly dense reading, but the most pertinent section is sec. 578, rem. 61. I will quote portions of it below.
Temporal clauses determining the duration of another action by the
duration of their own action, answering to the question 'how long,'
generally require both predicates to be in the PERFECT. They
correspond to English clauses with 'as (so) long as' (or 'while'
if it may be replaced by 'as long as'), and are introduced by the
conjunctions quamdiu, dum, or quoad.
Some examples Fischer gives:
Hortensius vixit tamdiu quam (or quamdiu) licuit in civitate
bene beateque vivere, Hortensius lived as long as it was possible to
live honorably and happily in the commonwealth. Cic. Brut. 1.4.
Ego dum vis fuit nihil egi; As long as force was reigning, I
did not do anything. Cic. Sest. 60. 127.
Gerd Haverling offers a complementary viewpoint in "Actionality, Tense, and Viewpoint" in New Perspectives on Historical Latin Syntax, vol. 2.
With expressions meaning 'while' or 'for the time that', Latin uses
the perfect to give a general overview ... and the imperfect to
indicate backgrounding.... The Latin for 'as long as he lived' is dum
vixit ..., and the imperfect is only used when backgrounding of the
situation is explicitly emphasized.... A similar distribution between
the perfect ... and the imperfect ... is found in Latin with
Up to this point, we've seen only past tense uses, but the general idea is that the perfect tense is the default (Fischer) unless the information is specifically intended as background information (Haverling). The situation is somewhat complex in cases of future time.
Haverling notes that in expressions of the sort 'if he lives' and 'as long as he lives', both the simple future and the future perfect are found, even when there is no clear anteriority (what Haverling calls "the absolute-relative sense"). He believes there may be a diachronic component to this:
In the function 'as long as X lives', Early and Classical Latin seems
to prefer the simple future ...; when Cicero uses the future perfect
of this verb, the absolute-relative sense seems always quite clear....
However, in a passage from the later Classical period and in passages
from the first centuries CE, the future perfect appears in expressions
meaning 'as long as X lives' ..., just as in Petronius.
Other examples of the development toward a more extensive use of the
future perfect are provided by the forms uoluero, which is never used
in Early Latin but which is met with in passages from the early first
century onward (e.g., Sen. contr. 10,2,1), and potuero, which may
easily have been confused with potero in the manuscript tradition, but
which nevertheless seems to become common first from the Classical
period onward (e.g., (165c) above). There are thus some changes in the
use of the future perfect from Early to later Latin, and in certain
cases the use of the future perfect becomes more frequent.
The problem in these cases concerns the extent to which the future
perfect in some passages refers to absolute-relative time, and to what
extent it has to do with an older aspectual system. There seems to be
a degree of overlap between the simple future and the future perfect
even when the same kind of actionality is expressed, and even in the
same verb. There are also some differences between different authors
as regards the use of one form or the other. The tendency to use the
future perfect in an absolute-relative sense is particularly strong in
Classical Latin, but the system is not totally consistent even there.
In my view, this indicates that in Early Latin we sometimes have to
reckon with some influence from an earlier system in which the
absolute-relative function was not yet as clear as it would be in
Classical Latin, and that it is likely that some such influence lived
on in fixed expressions in the colloquial language even during the
Classical and post-Classical periods.
Useful information, to be sure, but no definitive answer as to why a given author chose precisely the construction under consideration. My tentative suggestion is that the perfective aspect of the construction when used in the past may have been felt strongly enough to cause some authors to regard the future perfect as the more natural form.