Provide support for fields in abstract skeleton classes

[khatchad]

The refactoring should also provide support for fields in abstract skeleton classes, because fields cannot be easily migrated to a skeleton interface. Maybe fields could be migrated by adding getters and setters for each field to the interface and replacing field accesses by calls to these methods?

[khatchad]

This sounds like excellent future work or at least something to consider now.

[khatchad]

Might be related to the "method stubs" used in the Pull Up refactoring and/or #87.

[khatchad]

We may need to do this for the initial release, depending on how relevant the problem is. But, this really then becomes a composite refactoring (field encapsulation and then migration).

[khatchad]

Maybe our refactoring approach can invoke the encapsulate field refactoring (depending on if there are reads and/or writes) and then migrate those methods? Sounds quite complicated and I'm not sure if we'll have time for this.

[khatchad]

Putting in the future milestone now but may need to revisit.