I too am not a lawyer. But my understanding is this:
1. Anyone can look at Bronstein's pmint,
http://www-sop.inria.fr/cafe/Manuel.Bronstein/pmint/pmint.txt
Bronstein says
"pmint is not meant as a replacement for existing integrators, but either as
an extension, or as a cheap and powerful alternative for any computer
algebra project."
I think it is clear that he means cheap = 'easily implemented' not 'low in
monetary pricing'
2. Bronstein explicitly says in
http://www-sop.inria.fr/cafe/Manuel.Bronstein/pmint/
that
"its very small size makes it easy to port to any computer algebra system"
3. It appears that the python symbolic library has used it as the basis of
free code
http://code.google.com/p/sympy/issues/detail?id=463
4. And Axiom/Fricas, too
http://sage.math.washington.edu/home/page/fricas-src/src/input/pmint.input.p
amphlet
and maybe mathematica
http://compalg.jinr.ru/Dubna2008/abstracts.html
5. I do not know of any instance of Bronstein or INRIA selling research code
(but frankly I have not looked).
6. INRIA apparently was asked but did not respond to questions on this(?)
7. It may be the case that fair use of an electronically published free
program includes running or translating it, but even if not ...
To exclude the Maxima code for PMINT from the share directory when it is in
maple, sympy, axiom, fricas, sage... seems to me to be an excess of
caution. If you put it in the share directory and someone complains,
you can take it out. The "horse is out of the barn" in any case, because
INRIA has posted such code itself.
Richard Fateman
(incidentally, 2nd reader on Bronstein's PhD thesis).