Error handling

[Radiergummi]

Error handling should be a major priority. Each step has a wide range of possible errors that need to be handled differently:

• Reader errors are mostly fatal (No docs if sources are not readable), possibly warnings.
• Parser errors can be warnings (bad source, bad doc comments etc.) or fatal (Parser fails).
• Transformer errors should be warnings (non-transformable documents) mostly.
• Writer errors are mostly fatal (Cannot write, bad credentials etc.), possibly warnings too.

Especially parser errors are critical, since that touches code quality and logging too. We'll need a concept for handling those separately, maybe using user-defined error reporters.

[Radiergummi]

Note to self - some kind of IoC container might be helpful to establish control here, otherwise we'll resort to events.