Incorporation of variables for Data types

[JulStraus]

This is a first take at introducing variables as outlined in #11. It is a slight modification compared to the original proposal from my end to allow for declaring variables as well on the abstract type level. In addition, the filtering is used to avoid that a user must create two new functions. Instead, the design is similar to creating new variables for an element.

All data variables are created for a subtype of AbstractElement. This implies that functions must be created for each individual subtype. This is in line with the existing philosophy.

I included the approach as well for InvData to illustrate the application.

The PR also includes a renaming of Data to ExtensionData as proposed in #11. If undesired, we can revert the commit. Note that it is hence better to look directly at the individual commits.

Deprecated versions left within the code base

The following functionality is retained, but will be removed in a breaking release:

1. I left the old Data.
2. I left the call for variables_capex in create_model as well as the functions for Link and Node.