Add MinAbsolute to complement MaxAbsolute

[XapaJIaMnu]

This PR does two things: ~1) Changes the standard to C++17. Marian already uses that, there's no reason why we should continue with 11. gcc 5 supports almost the full C++17 anyways.~ 2) Adds MinAbsolute to complement MaxAbsolute. The purpose of this is to help me find which matrices will have shitty quantisation performance.