TODOs for 2.0

[Zaharid]

• Implement a proper set of unit tests, including those based on running the examples.
• Partially rewrite applwrap, in particular separating the LHAPDF and APPlgrid parts. Maybe base implementation on SWIG. It doesn't look like we can do much about the global state in any case.
• Make DB keys based on file hashes rather than filenames. This is easy but backwards incompatible. Also, for PDFs, we should really hash all the folder, which could be slow...
• Transpose the representation of results. This requires a lot of unittests but would make pandas work more naturally.
• Simplify the various intermediate steps to generate results. As long as produce_results is kept the same.
• Have default app folders, so that things like --use-db and data/ do not depend necessarily on the current dir in which smpdf is called.
• Add high level interface for scale variations.