[ID02] Error propagation

[kif]

• [x] Initialize variance array from user input using numexpr
• [ ] Propagate errors (as standard deviation, i.e. sqrt of variance)

[kif]

The performances are pretty good, see Variance.pdf

[kif]

Numexpr is 7.6 faster than numpy.