ENH: rename singlediode_methods.py to singlediode.py and discuss pvsystem.singlediode refactor

[mikofski]

Problem
Rename module name singlediode_methods.py and refactor pvsystem.singlediode

• singlediode_methods.py is a long name for a module
• the suffix "methods" is ambiguous, superfluous, and not descriptive
• the other modules have meaningful names that correspond loosely to the modeling steps described at Sandia PVPMC
• there needs to be a discussion regarding possibly refactoring pvsystem.singlediode
  • having a module shadow a method from a different module might be confusing or maybe it's not a problem?
  • has there already been discussion on breaking singlediode up, refactoring, or renaming it?

Proposed Solution
Rename module singlediode_methods.py to just singlediode.py and consider refactoring pvsystem.singlediode

1. rename the module from singlediode_methods.py to just singlediode.py
2. change all references in the code where appropriate
3. change all references in the comments and documentation where appropriate
4. start a discussion on the repercussions to the singlediode() methods in pvsystem.py

alternatives

• rename the module something else like dc_energy.py, pvmodules.py, or pvcells.py
• do nothing
• rename the module to singlediode.py and leave pvsystem.singlediode() alone
• rename the module to singlediode.py and move pvsystem.singlediode() to singlediode.singlediode() which is a more established pattern, right?

Additional context

• this is a follow on to ENH: implementing pvsyst recombination loss current for CdTe and a:Si #504
• comments in move lambertw to singlediode-methods module #497
• also related to singlediode is slow #408, ENH: add methods and tests for a explicit IV curve calculation of single-diode model #409, and Single diode model helper function to wrap Lambert w and other methods #410

[cwhanse]

My mental model has the following use cases:

• calculate IV curve parameters for a specific single diode model at given irradiance and temperature conditions. This is the job of calcparams_XXXX, there's no way to avoid multiple functions because the single diode models use different sets of equations.
• calculate an IV curve for a given set of IV curve parameters. This is the job of singlediode which should accept a kwarg to select a calculation method, e.g., lambertw, bishop88, others perhaps @thunderfish24. The standard single diode equation (SDE) and SDE+recombination are different. If possible, we should handle calculations for the SDE and SDE+recombination within a IV curve method: bishop88 can and is the subject of ENH: implementing pvsyst recombination loss current for CdTe and a:Si #504. Extending other methods to accommodate the recombination term is requested by this issue. However, I already know that the solution technique used in the lambertw method cannot be extended to SDE+recombination.
• ModelChain should have a method that conveniently wraps the various DC models, including sapm, Desoto, CEC, PVsyst. We have a partial implementation in ModelChain.dc_model.

Regarding #517, I vote strongly against singlediode_pvsyst unless we can't find any other workable way to host the several single diode models.

[wholmgren]

I suggest we rename singlediode_methods.py to singlediode.py as soon as possible. I think this is a blocker for 0.6 but the rest of it could be pushed off to 0.7.

[cwhanse]

Ok.

[mikofski]

FYI: I'm on it right now, and BTW: Cliff was right, they clashed in pvsystem.py, but it's no problem.