Excessive warnings about Nxsec vs Nweights mismatch
Cf. lots of email and ATLAS discussion, the warning message (every event) about mismatching numbers of weights and cross-section entries is overzealous, and the high verbosity is making HepMC unusable in many contexts. While the two ideally should match, there is realistically no problem when there is only one cross-section, as is very common: the variation cross-sections can be (and are) computed in user code by multiplying the one (nominal) cross-section by the ratio of weight-sums in the run.
We should avoid warning in that case for now -- and if we do later introduce a warning for single cross-sections on multiple weights (I think it does mean that the xsec error can't be computed exactly), suppress it to a finite number of printouts and deal with the resulting global state or thread locality.