A-posteriori check of ROQ on h_+*h_x terms

We construct a basis using only h_+ terms, under the hypothesis that h_x ones will have comparable accuracy.
However, this should be checked a-posteriori, as in arXiv:160408253 Sec. III.E (see also https://arxiv.org/abs/1604.08253).

Since for the 22-only case this term is exactly zero, for the moment, hphc terms have been commented out to ignore ugly Runtime warnings. Should be called only when are non-zero (HMs etc.).