Remove rsh = -dV/dI @ Isc constraint in gen_coeffs

[mikofski]

I have used this constraint successfully for years, but it's time to let this go

1. It is too restrictive. It fits well SunPower modules, but not other modules and also not other irradiance conditions
2. Other fitting algorithms do not use it although some do, maybe in the docs or wiki we can do a review. SAM does not use this constraint, instead they use the point at short circuit
3. There is only weak theoretical bases for this constraint, but mostly, it's inconsistent with how we generate IV curves - should the fitting procedure just be the opposite of the calculation?

I propose we follow SAM and just use these four

1. Point at short circuit
2. Point at open circuit
3. Point at mpp
4. slope at mpp of power-current cure should be zero

We already use 2, 3, and 4, and adding 1 is trivial, it's actually already in the because it's use to calculate Igen (aka Iph or also IL)