Use channelid in cudacpp (and fortran) helicity filtering?

[valassi]

Use channelid in cudacpp helicity filtering?

Hi @oliviermattelaer this is another followup to our chat yesterday. You suggested that we should use channelid in helicity filtering in cudacpp, whicl no wwe do not. Essentially you suggest to remove the nochannel calls to ME calculations.

The point is that strictly speaking the list of good helicities depends on the channel. Rephrase: a channel has ME>0 only if the numerator (one individual feynman diagram) is greater than 0. So there may be helicities which are good for one diagram, but always give 0 for another diagram.

IIUC however this is something that is not yet done in fortran? Or rather, IIUC, you said that there is a single list of helicities common to all channels n fortran, which is computed by iterating over all channels, so that a "good helicity" is a helicity which gives ME>) for at least one channel. This is essentially what cuudacpp also does now.

This is a followup to #960 (where you noticed that there is a FBRIDGESEQUENCE_NOMULTICHANNEL call). Anyway we agreed to merge #960 as is, and then eventually work on LIMHEL>0 in cudacpp (#564) and on helicity filtering with channelid (this issue)