add BESA .elp support to read_montage

[jona-sassenhagen]

Direct translation of EEGLAB routines for reading spherical BESA .elp channel location files.
Not to be confused with polhemus .elp files (no overlap with #1001).
Possibly need smarter way of handling header files; the only example I know has 1 header line, but there may be other formats.

Example elp file: https://sccn.ucsd.edu/svn/software/eeglab/plugins/dipfit2.3/standard_BESA/standard-10-5-cap385.elp

[agramfort]

please run pep8 checker on this file.

Also if numbers are floats don't forget the . so 90 becomes 90. (180 -> 180.) etc.

[agramfort]

you'll need to add a test for this. See how it was done for other montage files.

https://github.com/mne-tools/mne-python/blob/master/mne/channels/tests/test_montage.py

Run the tests with

nosetests mne/channels/tests/test_montage.py

thanks for the cool contrib!

[jona-sassenhagen]

On it!

[agramfort]

Can you share a valid elp file?

[jona-sassenhagen]

I have corrected PEP8 errors, added dots to mark the numbers as float and updated test_montage.py.
The test runs "ok" on my machine.

The .elp file I'm using is the one linked above: https://sccn.ucsd.edu/svn/software/eeglab/plugins/dipfit2.3/standard_BESA/standard-10-5-cap385.elp
It's the one packaged with EEGLAB (which is GPL).

I've also re-added the conversion to string for .sfp files.

Is this okay?

[agramfort]

merged by rebase after pep8 fixes, adding your name to file + what's new page

b26ab1a
dd0009a

thanks @jona-sassenhagen !