The benefit of disposableness is, well, that data will be disposed of.
Why might you want to do this using an appVM?
Lets say you have a template which provides firefox, and you configure a
couple of qubes using that template - one has ad-block, and other
security oriented plug-ins, and a selection of your favourite bookmarks;
the other has no plugins, but a few bookmarks.
So the obvious advantage of basing disposableVMS on the qubes is that you
have firefox configured just the way you like and access to relevant
bookmarks, but you have the relative security of knowing that pretty
much anything hostile will disappear when you close the browser, and the
disposableVM is cleared.
You could do this in the template but it would be somewhat
more difficult (and risky) since you would have to give the template
network access to install plugins. Of course, you'd also need two
templates to provide the different needs of the qubes.
Naturally, all disposableVMs based on a qube will carry the same
fingerprint as the qube, so you cant use them to provide a measure
of separation between identities. (The same goes, to some extent, to
qubes sharing a template.)
In your case, there may not be any advantage in using a a qube as
opposed to a template. I don't know what you are trying to achieve in
using Qubes.
You need to look at the capabilities and decide what will best suit your
needs.
unman