[kamillaurent]

For $Q_0 < m_c$,construct_assumptions returned Nf4Sym, which drops V15 (c-) from all FK tables. Now it returns Nf4Ind (as in $Q_0 > m_c$). I also updated the test test_theory_card.py.

I checked that the fix works using theory 41002000 (4.1, $Q_0 = 1 GeV$)$ for computing the FKs for ATLAS_1JET_8TEV_R06_PTY_BIN1: pineko logs Optimizing for Nf4Ind.

[felixhekhorn]

The code looks good. Please add a Changelog entry. To echo what we just discussed: we need the LHAPDF fixes for the benchmarks most likely