Clean up physics parameters across CPPProcess and mgOnGpuConfig

[valassi]

Clean up physics parameters across CPPProcess and mgOnGpuConfig

The physics parameters like number of initial or final particles etc were meant to be in CPPProcess (and are still in the header, unused). But they are now in mgOnGpuConfig (I am the culprit who put them there). They should go back to CPPProcess.

Note also that they should all be constexpr (issue #265)

Cleanup for after epochX3

[valassi]

Note that this is spot-on the issue we face when we have two P1 subdirectories, as in the pp to 4top of #534 and MR #560

[valassi]

Specifically for the pp to 4top, the differences I get between pp to 4top and qq to 4top are the following

diff -r gg_tttt.sa/src/mgOnGpuConfig.h qq_tttt.sa/src/mgOnGpuConfig.h
129c129
<   const int nwf = 18; // #wavefunctions = #external (npar) + #internal: e.g. 5 for e+ e- -> mu+ mu- (1 internal is gamma or Z)
---
>   const int nwf = 13; // #wavefunctions = #external (npar) + #internal: e.g. 5 for e+ e- -> mu+ mu- (1 internal is gamma or Z)


diff -r gg_tttt.sa/src/Parameters_sm.cc qq_tttt.sa/src/Parameters_sm.cc
103d102
<   GC_12 = mdl_complexi * mdl_G__exp__2;
188d186
<   std::cout << std::setw( 20 ) << "GC_12 = " << std::setiosflags( std::ios::scientific ) << std::setw( 10 ) << GC_12 << std::endl;


diff -r gg_tttt.sa/src/Parameters_sm.h qq_tttt.sa/src/Parameters_sm.h
43c43
<   //cxsmpl<double> GC_10, GC_11, GC_12; // now computed event-by-event (running alphas #373)
---
>   //cxsmpl<double> GC_10, GC_11; // now computed event-by-event (running alphas #373)
176d175
<   //constexpr cxsmpl<double> GC_12 = mdl_complexi * mdl_G__exp__2; // now computed event-by-event (running alphas #373)
197c196
<   constexpr size_t ndcoup = 3; // #couplings that vary event by event because they depend on the running alphas QCD
---
>   constexpr size_t ndcoup = 2; // #couplings that vary event by event because they depend on the running alphas QCD
200d198
<   constexpr size_t idcoup_GC_12 = 2;
205d202
<     cxtype_sv GC_12;
233d229
<       out.GC_12 = cI * mdl_G__exp__2;
276d271
<     fptype* GC_12s = C_ACCESS::idcoupAccessBuffer( couplings, idcoup_GC_12 );
279d273
<     cxtype_sv_ref GC_12s_sv = C_ACCESS::kernelAccess( GC_12s );
282d275
<     GC_12s_sv = couplings_sv.GC_12;

In principle, the only real bug here is the value of nwf, as suggested by @zeniheisser . The fact that we have 3 parameters instead of 2 in the pp process is instead not an issue, provided that the indexes for these parameters are propagated correctly

[roiser]

I can confirm that nwf is the only parameter that needs to move "down" into the P1 directories.