Create a prototype of a "reduced" engine facade

**Is your feature request related to a problem? Please describe.**
This is an idea for different way of dealing with the complexity of the force field definition.

The current implementation of MDMC constructs all the MD engine input from scratch. We have been considering the "black box" approach as the other extreme: in this case MDMC would take the input files and modify them without understanding the contents, and calculate the observables from the resulting trajectory.

Following the discussion with the DL_POLY authors, we have created a draft of a compromise solution, a reduced engine facade.

**Describe the solution you'd like**
The main features of the reduced engine facade:
1. It is tied to a specific MD engine (e.g. LAMMPS, DL_POLY),
2. It does **not** create all the input files, but it parses and validates all of them.
3. The force field definition is created by the *user*. Placeholders are put in the file for the parameters that will be refined.
4. The starting structure is created by the *user*. MDMC validates the structure, checks if the box size, number of atoms, etc. are sufficient.
5. The engine script/control file is created by *MDMC*. This way MDMC retains control over the number of steps, step size, type of the run, etc.
6. The method for loading the trajectory after the run is chosen to match the MD engine output, same as it is the case for our existing facades.

**Describe alternatives you've considered**
The "black box" approach, in which ALL the input files are provided by the user, makes it easy to start a simulation for any engine, but quickly runs into difficulties when the output trajectory has to be loaded.

**Additional context**
The main assumption is still that any user who wants to *refine* a model of their system must have already run a simulation, and therefore has all the inputs needed to run another one.