request for differentiation stress testing

[Chris Smith]

In PR15049  (the smichr/diffsort branch) I have made modifications to differentiation related code that I would like to see put through more extensive testing. There are already many issues fixes, especially as related to higher order derivatives, but I would like to know about remaining issues that might show up from work that people are already doing (and already know that it doesn't work in the master branch). You can post issues on the PR itself rather than here.

Changes there affect how doit responds, how subs behaves and the canonical ordering of arguments when differentiating wrt more than one object.

Grateful for comments already from @jksuom and @asmeurer.

Thanks in advance,

/c