convert 1D shape into largest perfect md shape possible

[Christian-B]

Simplest version uses a m D shape with all sides the same size.

So 256 in 3D would be (6,6,6) where a shape of (7,6,6) would be possible

[andrewgait]

No objections as long as it doesn't break any current tests.

[andrewgait]

Though perhaps a test could also be added using the script at http://spinnakermanchester.github.io/development/2d_convolutions.html

[Christian-B]

This is only one of several options.

Another is to keep as master and require the setting of max atoms.

Another is to have 2D defaults

[andrewgait]

I don't know how this was resolved so leaving it open and re-assigning it.

[coveralls]

coverage: 83.951% (+0.3%) from 83.65%
when pulling 3ee989d on md_default
into b9c1c42 on master.

[Christian-B]

Assume this is no longer needed.

[rowleya]

Yes, I think this was a test on the path, but is now no longer needed.