Robust algorithm for 2-colorability

[siraben]

Needed so we can produce a novel 2-coloring for the neighborhood of a vertex in a 3-colorable graph. Robustness is needed so we do not need to know anything a priori about the 2-colorability of a graph and can produce either a complete 2-coloring or a proof that the graph is not 2-colorable.

Maybe it would be cleaner to construct the witness (odd cycle) in the negative case but it would be more algorithmically involved.