Abstract config reader functionality for `train` schema

[CBroz1]

The following was provided as feedback

A lot of functionality could be abstracted as it is not DJ specific. For example config parsing is currently done within a schema (example) while an own class could be the better choice here. This could also robustify the code, help with versioning, etc. For example, say a lab uses a non-standard DLC file, how would I be able to use elements w/o actually changing the code? If config parsing is in its only module, features like this could be very easy implemented.

[CBroz1]

This would likely not involve any new features, but rater a refactoring of existing code and demonstrating coverage via existing pytests