Variant integration in simulation engines

[Luthaf]

Now that we have variants for outputs (#91), we should integrate them in the various simulation engines:

• plumed should support features/one and features/two
• chemicope should also support features variants
• LAMMPS should support variants on the energy, non-conservative forces and stress, and energy uncertainty
• TorchSim should support variant for the energy
• eOn should support variant for the energy
• i-PI should support variants on the energy, non-conservative forces and stress, and energy uncertainty

[ceriottm]

Is this already exposed by lj_test? I can take care of i-PI.

[Luthaf]

Yes, there is now a /doubled variant for all relevant outputs

[PicoCentauri]

I will take care of the lammps interface