Improve error handling for missing input genomes and reference databases.

[njohner]

With this PR we add a process to the beginning of the annotation pipeline to check the presence of the necessary reference databases. We also ensure that all the input genome files exist and are accessible. For both cases we raise specific errors that should make it clear what the issue is. This was not the case before with errors that could be rather cryptic.

Closes #128

Checklist

• Changelog entry
• Check that tests still pass
• Add tests for new features and regression tests for bugfixes whenever possible.