Implement reader for REDUCE files, superceeds #14

[hamogu]

As suggested by @wkerzendorf, I split the old pull request #14. This is the REDUCE part.

[hamogu]

@wkerzendorf Since the master has changed again after merging #20, I would have to rebase and modify my readers (This and issues #16 and #18) again. However, before I do that I would like your opinion about them - if you think we should not have these capabilities yet because the Spectrum1D API is not stable yet, then I'll save that time.
I already invested some time in rebasing them after you asked me to split #14 into separate pull requests; I fully understand if you want to wait until after the API document is there, but if that is the case then I won't spend any time on them either.

[wkerzendorf]

@hamogu: I'm currently working on the API (as in right now) - wait a little bit. I should have something by this afternoon/evening.

[astrofrog]

This should be a relative import

[hamogu]

Unfortunately, this will need to wait a little longer. In the current implementation, dispersion and flux can no longer be 2-dim arrays. We'll need a Spectra1DEchelle object or something like that. That's discussed in depth in #11, and will probably only be decided after we have an API document out there.

[wkerzendorf]

Superseded by @shaileshahuja implementation