DISC fixes

[wholmgren]

Note the small API changes.

Could be some merge issues with #22. I'll take care of that if it happens.

[wholmgren]

Here's some sample output. Not sure what the cause of the errors in the first diagram is since I haven't used this algorithm very much. Maybe airmass and clear sky model mismatches? I did check the Python code against PVLIB_MATLAB 1.2 and it looks ok.

tus = Location(32.2, -111, 'MST', 700)
tus_pressure = pvlib.atmosphere.alt2pres(700)
times = pd.date_range(start=datetime.datetime(2014,6,24), end=datetime.datetime(2014,6,25), freq='1Min')
solpos = pvlib.solarposition.get_solarposition(times, tus, method='pyephem')
tus_cs = pvlib.clearsky.ineichen(times, tus)
ax = tus_cs.plot(figsize=(8,6))
DNI = pvlib.clearsky.disc(tus_cs.GHI, solpos.zenith, solpos.index, pressure=tus_pressure)
DNI.DNI_gen_DISC.plot(ax=ax, legend=True)

tus = Location(32.2, -111, 'MST', 700)
tus_pressure = pvlib.atmosphere.alt2pres(700)
times = pd.date_range(start=datetime.datetime(2014,1,24), end=datetime.datetime(2014,1,25), freq='1Min')
solpos = pvlib.solarposition.get_solarposition(times, tus, method='pyephem')
tus_cs = pvlib.clearsky.ineichen(times, tus)
ax = tus_cs.plot(figsize=(8,6))
DNI = pvlib.clearsky.disc(tus_cs.GHI, solpos.zenith, solpos.index, pressure=tus_pressure)
DNI.DNI_gen_DISC.plot(ax=ax, legend=True)

[robwandrews]

Interesting, it looks like the algorithm is good, but there might be some science to look at the difference in the algorithms. Looks good for a merge.