Add xyz writer

[mattwthompson]

Since XYZ files store next to nothing (no residue information, box, bonds, angles, dihedrals) I was able to just enumerate through the sites and avoid the complications Ray is having in #28.

I added an element setter and getter, which may or may or may not affect other behavior but seems like something we should have already done.

The logic in assigning a name could be improved, I just check to see if there's an element stored in the site and if not, I try to assign a "name" according to the first characters of the site's name, and if those fail I just call it X.

[justinGilmer]

What about the case when a name is >2? Wouldnt we want to support that as well?

Even though its not an element anymore, i could see a case where your sites are not atomistic representations.

[mattwthompson]

This column in XYZ files is supposed to be element. We need better element guessing, but I don't think there are good rules for handling non-elements in this format. For now I guess we can just let the entire name be stored, but it may be problematic down the line - we'll see.

[justinGilmer]

date would be nice to have

[ahy3nz]

Further down the line, we might want to do some validation of the element

[ahy3nz]

type checking?

[mattwthompson]

I would prefer doing this in a separate PR bulking up the element class

[mattwthompson]

I added a simple datetime printer, the result looks like this:

8
Topology tmp.xyz written by topology at 2019-02-24 10:52:34.097767
X    0.000   -1.400   -0.000
X   -1.070   -1.400   -0.000
X    0.357   -2.169    0.653
X    0.357   -1.581   -0.993
X    0.000    0.000    0.000
X    1.070    0.000    0.000
X   -0.357    0.769    0.653
X   -0.357    0.181   -0.993