Feat/performance ratio

[kandersolar]

References #127. Step one of the bigger loss categorization project is to add the ability to calculate performance ratio. The method chosen for now is the temperature-adjusted metric defined in [1]. I've put the code in a new file called losses.py in anticipation of future loss-quantification functions going there as well.

The main user-facing function performance_ratio differs from the spec in two ways:

• optionally allowing the user to filter out times where the expected power would be above the system's clipping limit
• defaulting to 25 C as the reference cell temperature instead of the POA-weighted Tcell value calculated from a system's 8760. The POA-weighted Tcell version is intended to give the same annual value as the standard (non-temp-adjusted) PR, which results in the values being centered around 0.8 or so depending on the system. Using 25 as the reference shifts the values to centered around 1.0.

The other function calculate_pr is more generic and allows a user to calculate a PR using a generic expected energy signal and filtering mask.

[1] "Weather-Corrected Performance Ratio" - https://www.nrel.gov/docs/fy13osti/57991.pdf

[kandersolar]

Current Travis failures are from the py2.7 jobs, FYI

[mdeceglie]

Closing this one, it's been stale for a while