add single axis tracking

[wholmgren]

This PR contains a Python implementation of PVLIB MATLAB's pvl_singleaxis.m. This isn't yet ready to merge (needs tests and PEP8), but I wanted to share progress and provide a place for feedback.

The code and comments are very similar to pvl_singleaxis.m. I think that some things can be simplified, but I wanted to get something working first. Sandia put a lot of comments in the code and I'm inclined to keep most of them.

I believe that the AOI calculation is correct, but there's something wrong with the surface azimuth and tilt calculation.

Here's an IPython notebook that explores most of the parameters.

[wholmgren]

A lot of changes since the initial PR.

• Fixed the problem with surface_tilt and surface_azimuth. I'm not sure whether the MATLAB authors intended for surface_tilt to be with respect to surface horizontal plane or surface normal.
• Parameters and most of the inline code has been PEP8-ified.
• The MATLAB authors required a latitude input and threw an error if the latitude was less than 0, but I think the new code works fine independent of the latitude.
• The sample notebook walks through each step of the algorithm and then tests the whole thing in many regimes.
• I added some unit tests.

It would be helpful for somebody to make an initial review at this point.

[alorenzo175]

What if the given elevation and azimuth don't have the same index? Could run into problems when doing things like finding wid. I would suggest ensuring/forcing apparent_zenith and apparent_azimuth to have the same index that you then use for times.

[wholmgren]

@alorenzo175 pointed out that the surface_tilt calculation was done wrong, and that surface_azimuth did not work for non-integer multiples of 90 degrees. I think I've fixed both problems in the latest commit, and added some explanatory comments in the code and notebook. The second half of the surface_azimuth calculation is now significantly different than that of the MATLAB code.

I expect that the tests will fail since I didn't update the expected values. I'll make the other changes too.

[wholmgren]

Updated the tests and added a few lines for @alorenzo175. Added introduction material to notebook.

[wholmgren]

Added the same examples as the PVLIB MATLAB 1.2 help file to the tutorial. Updated internal variable names to be consistent with PEP8 and, I think, make the algorithm more understandable. Also replaced lots of arctan mapping code with arctan2.

I think this is close to being ready to merge.

[wholmgren]

Rebased so that the test run against all of the pandas versions.

Removed the latitude kwarg since I don't think it's necessary.

I am happy to make changes, but if there are no objections to merging then let's go ahead and do it.