Please continue to post to the list, not to me personally,
so others can answer with ideas.
1) I don't know if an NVT thermostat will be enough - you hvae
to be careful about which atoms you thermostat and what degrees
of freedom. Since you are pulling on some atoms and some
are constrained in some dims, applying NVT to group all is probably not
a good idea. See section 4.16 of the docs for thermostatting
ideas. Note that you also need to be careful about how you
measure temperature when some atoms are constrained.
2) You can do either of the ideas you suggest - LAMMPS is very
clear about what it is outputting in both cases. Which one corresponds
to what you want to measure is really up to you to figure out.