Include overwrite data option in SimulationDataSource

[flongford]

This PR introduces an option on the SimulationDataSourceModel class to overwrite any simulation data if it already exists on file. Consequently, a new check is performed before running the simulation to see whether

• A: an output data file already exists
• B: the SimulationDataSourceModel requires this data to be overwritten

If A returns False, then a simulation is always performed.

If A returns True, then a simulation will only be performed when B returns True also.

Whilst implementing this check, we have refactored some of the required functionalities of GromacsSimulationBuilder into the new interface class ISimualtionBuilder. This can be considered a step towards obataining generic simulation classes for the BDSS, similar to #14.

Change log:

• SimulationDataSourceModel class includes ow_data attribute
• New ISimulationBuilder interface class requires subclasses to implement a get_results_path method, which will define output file location for simulation data.
• GromacsSimulationBuilder class now provides ISimulationBuilder class, and so can be queried to provide an output data file path before a simulation is performed.
• SimulationDataSource class includes context manager block using _check_perform_simulation method to return outcome of A and B checks for an GromacsSimulationBuilder instance
• Plugin no longer contributes SimulationFactory to be usable from the BDSS, since the SimulationDataSource requires the create_simulation_builder method to be implemented

[codecov-io]

Codecov Report

Merging #16 into master will increase coverage by 0.13%. The diff coverage is 96.77%.

@@            Coverage Diff             @@
##           master      #16      +/-   ##
==========================================
+ Coverage   94.55%   94.68%   +0.13%     
==========================================
  Files          34       35       +1     
  Lines         826      847      +21     
  Branches       98      101       +3     
==========================================
+ Hits          781      802      +21     
  Misses         31       31              
  Partials       14       14

Impacted Files	Coverage Δ
force_gromacs/gromacs_plugin.py	100% <ø> (ø)	:arrow_up:
force_gromacs/api.py	0% <0%> (ø)	:arrow_up:
...romacs/data_sources/simulation/simulation_model.py	96% <100%> (+0.16%)	:arrow_up:
...romacs/pipeline/base_gromacs_simulation_builder.py	96.15% <100%> (ø)
force_gromacs/pipeline/i_simulation_builder.py	100% <100%> (ø)
.../data_sources/simulation/simulation_data_source.py	96.87% <100%> (+1.04%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 20e4453...55b306c. Read the comment docs.

[flongford]

Happy to merge then, if you think we want to keep some methods as NotImplemented for now.

Sure - similar to some objects in the force-bdss, the intention is for these objects to act as base classes for developers to implement. Do you think therefore it would be clearer to rename the class containing these methods to BaseGromacsSimulationBuilder?

[Corwinpro]

Happy to merge then, if you think we want to keep some methods as NotImplemented for now.

Sure - similar to some objects in the force-bdss, the intention is for these objects to act as base classes for developers to implement. Do you think therefore it would be clearer to rename the class containing these methods to BaseGromacsSimulationBuilder?

Good question, I didn't think about it - I thought that the GromacsSimulationBuilder is a ready to use class. How different can the NotImplemented methods be?

[flongford]

Happy to merge then, if you think we want to keep some methods as NotImplemented for now.

Sure - similar to some objects in the force-bdss, the intention is for these objects to act as base classes for developers to implement. Do you think therefore it would be clearer to rename the class containing these methods to BaseGromacsSimulationBuilder?

Good question, I didn't think about it - I thought that the GromacsSimulationBuilder is a ready to use class. How different can the NotImplemented methods be?

They have a fixed input / output signature, but otherwise are pretty generic - we could probably think of a better way to handle this using defaults but that might be out of the scope of the current PR