Add LHCbMath/Truncate.h that implements methods to truncate data

Add LHCbMath/Truncate.h that implements methods to truncate floating point values to a fixed number of d.p.

Derived from a method I currently have private to a RICH algorthm, but which I now wish to use in a number of places.

I have also improved it to support truncating SIMD values as well as basic scalars.

@cattanem ready to test.