Improve vec_utils functions

[jonblack]

Check if the following libraries can make life easier. This can be as little as providing more useful types for performing matrix/vector operations to as much as providing the implementation so we can delete what we've written:

• Boost.uBLAS
• Eigen
• NT2

The functions that have been written, if kept, should be templated so they are generic for all vector types. I'm actually hoping that the above libraries will provide something like numpy.matrix in python because my feeling is that using vectors will be really slow, so we may want to use raw arrays or something else.

Leave this issue until near the end. W can decide that later when we have everything tested and can run a profiler on the code to see where it's slow.

[joannalange]

vec_util functionality replaced with library functions