Port post-generation patches (patchMad.sh) upstream to mg5amcnlo or to the plugin

[valassi]

Presently code generation in cudacpp goes through the patchMad.sh script, currently:
https://github.com/madgraph5/madgraph4gpu/blob/3a4850184f1de483bbab1bff5a6a322590692291/epochX/cudacpp/CODEGEN/patchMad.sh

This applies post-generation patches to the code generated using the python madgraph machinery. For a release, these patches must be instead moved upstream to mg5amcnlo (or to the plugin itself).

The main ingredients of this are the following (I may be forgetting a few):

• add a few files to fortran code such as timer.h ompnumthreads.cc (these may best go to the plugin?)
• apply patches patch.P1 and patch.common (these have already been reduced significantly by the vector.inc MRs First set of vector.inc changes for cudacpp integration mg5amcnlo/mg5amcnlo#23, Second set of vector.inc changes for cudacpp integration (make VECSIZE_USED a function argument) mg5amcnlo/mg5amcnlo#24, Minor 'vecsize3' changes (improve passcuts interface and add comments about vector.inc) mg5amcnlo/mg5amcnlo#54: the rest should go partly to the plugin an dpartly maybe upstream, eg a general mechanism to patch driver.f must be implemented, maybe also for sycl)
• fix a bug in coupl_write.inc, probably this should go upstream (avoid writing n times the coupling if the couplking is a pointer not a value)
• also very important, generate coloramps.h from coloramps.inc

Assigning this to me, cc @roiser and @oliviermattelaer

[valassi]

Note, this is part of point 2 in #576

[valassi]

I have started a WIP patch for this in MR #675