Improve initialization of proteins and other rigid objects

[pm-blanco]

Request by @kosovan, the initialization of globular proteins could be improved:

• Currently, globular proteins are initialized at random positions, which can lead to overlaps for non-diluted solutions. We should implement a lattice-like initialization.
• By default, globular proteins are fixed in the simulation box and the user has to enable its motion using a built-in function in pyMBE. It would be more intuitive to do it the other way around, which is also the standard in Espresso

[pm-blanco]

@kosovan @paobtorres UPDATE on this topic after revisiting the features for globular proteins:

• I think that the current workflow of having a fixed globular protein as "default" and then forcing the user to enable its motion if needed is actually the way to go. Enabling the motion of the protein as a rigid objects requires additional setups (defining virtual sites, adding proper mass and moment of inertia to the central particule) that the fixed object does not.
• We still should work in better initialization strategies if we want to enable simulations with several globular proteins

[pm-blanco]

Solving this issue could become important for progress in #144