implemented LagrangeQuadratic order 2

[YetAnotherMinion]

The doxygen indicates that getLagrange for dimensions 1 and 2 are implemented, while in fact dimension 2 is not.

Get the Lagrangian shape function of some polynomial order.
we have only first and second order so far

I think I got the shape functions right, but please check. My motivation is to get apf::Mesh::changeShape(apf::getLagrange(2)) to work and automatically create the 9 nodes. I do see where the new nodes get added to the mesh, so if you can point that file out I can also address that too. I could not find the ordering of the nodes in eta space, so I went with ordering in Hughes. If the shape functions and gradients are correct, then it should be trivial to reorder them by someone who does know the ordering.

[YetAnotherMinion]

Also it makes the shape functions much easier to read when you substitute a single symbol for xi[0], xi[1] ,... It will get optimized out by compiler anyway and makes the tight grouping style of arithmetic operators favored by this code base a little easier to understand. If the shape functions look good one can always do a local find and replace to put those xi[i]'s back if that is the convention

[ibaned]

yea, theres a whole funny story about the serendipity quad but the short version is I was lazy and you seem to have wrapped it up nicely so I'll pull this and run a couple sanity tests.

Thanks !

by the way, the nodes will take care of themselves, that countNodes function controls it all.
also, the single-char names are fine, but its dangerous to go back and change the existing
code so I'll just take this new stuff as is.