Improve convergence of lambertw for very large arguments

[cwhanse]

No description provided.

[cwhanse]

The test failures don't look to be caused by my one edit which changes range(0,3) to range(9) in pvsystem.

[wholmgren]

Is it easy to add a test that requires this change in order to pass? test_pvsystem.py has the following function that you could copy/paste, rename, and then update with your parameters:

@requires_scipy
def test_v_from_i_bigger():
    # 1000 W/m^2 on a Canadian Solar 220M with 20 C ambient temp
    # github issue 225
    output = pvsystem.v_from_i(190, 1.065, 2.89, 0, 7.05196029e-08, 10.491262)
    assert_allclose(54.303958833791455, output, atol=1e-5)

For reference, this is a continuation of #268, #266, #260.

[cwhanse]

I haven't been able to reproduce the failure, so let's close this pull request. We can always re-open if we find the failure conditions again.