Efficient spherical averaging by use of symmetry

[ajjackson]

Depends on #85
Blocking #25

Use symmetry information to improve performance of S(q, ω) sampling over spherical surface #85

This would likely involve a dependency on spglib and/or Brille.

Requires: functionality in Euphonic to identify or pass on information regarding the spacegroup symmetry of a structure.

[ajjackson]

An outcome of #116 discussion was the observation that powder averaging could benefit from a functioning Euphonic-Brille interface, especially at long distances from Gamma, where many samples may be needed and might be cheaply obtained by interpolation from a mesh within the Brillouin-Zone.

As such, powder-averaging could be a good early case-study or application of the Euphonic/Brille interface.

[ajjackson]

On reflection, I am convinced that powder-averaging with symmetry-reduction is a bad idea. We are getting good results from quasirandom meshes of points that have equal weights. It doesn't cost more for Euphonic to compute the point inside or outside of the IBZ. So I think we can close this with prejudice until a different sampling approach comes along.