Start implemnting some basics on the new output handling.

Not much to be seen, yet. General ideas:
- Outputs are imported from their storage locations by copying their folder to a temp location, recursively.
- Directly after importing, a hash is created to detect whether an imported output needs saving.
- Saving is just the reverse: Copy the temp folder to the storage folder.
- Storage folder may be relative to workspace.
- On exit, check all outputs known in that session for changes, prompt for saving.
- Several output folders will be supported, whith actions to create, drop, activate.
- When dropping an output folder, another one is activated automatically (if only one, this corresponds to the current "flush output").