use _lambertw_pvlib etc.

[cwhanse]

• Closes #xxxx
• I am familiar with the contributing guidelines
• I attest that all AI-generated material has been vetted for accuracy and is in compliance with the pvlib license
• Tests added
• Updates entries in docs/sphinx/source/reference for API changes.
• Adds description and name entries in the appropriate "what's new" file in docs/sphinx/source/whatsnew for all changes. Includes link to the GitHub Issue with :issue:`num` or this Pull Request with :pull:`num`. Includes contributor name and/or GitHub username (link with :ghuser:`user`).
• New code is fully documented. Includes numpydoc compliant docstrings, examples, and comments where necessary.
• Pull request is nearly complete and ready for detailed review.
• Maintainer: Appropriate GitHub Labels (including remote-data) and Milestone are assigned to the Pull Request and linked Issue.

Replace calls to scipy.special.lambertw with ivtools.utils._lambertw_pvlib

[kandersolar]

sdm/_fit_deosoto_pvsyst_sandia.py and sdm/desoto.py use scipy.special.lambertw. Worth replacing there as well?

[kandersolar]

Some brief, noisy, non-comprehensive testing shows significant speed-ups. Here's the ratio of runtimes of PR and v0.13.1, for simulations of ~1e6:

singlediode    0.309317
i_from_v       1.068197
v_from_i       0.237070

[kandersolar]

Seems like this warning is commonly (always?) emitted now:

pvlib-python\pvlib\ivtools\utils.py:563: RuntimeWarning: invalid value encountered in subtract
  w = logx - np.log(logx)  # initial guess, w = log(x) - log(log(x))

Here's a reproducer:

import pvlib
cec = pvlib.pvsystem.retrieve_sam('cecmod')
sdm_params = cec['Canadian_Solar_Inc__CS1U_400MS']
sdm_params = sdm_params[['alpha_sc', 'a_ref', 'I_L_ref', 'I_o_ref', 'R_sh_ref', 'R_s', 'Adjust']]
sde_params = pvlib.pvsystem.calcparams_cec(1000, 25, **sdm_params)
iv = pvlib.pvsystem.singlediode(*sde_params)

[cwhanse]

Seems like this warning is commonly (always?) emitted now:

pvlib-python\pvlib\ivtools\utils.py:563: RuntimeWarning: invalid value encountered in subtract
  w = logx - np.log(logx)  # initial guess, w = log(x) - log(log(x))

This is happening because overflow arguments are passed into _lambertw_pvlib here:

        # Record indices where lambertw input overflowed
        idx_inf = np.isinf(argW)

        lambertwterm = _lambertw_pvlib(argW)

[echedey-ls]

sdm/_fit_deosoto_pvsyst_sandia.py and sdm/desoto.py use scipy.special.lambertw. Worth replacing there as well?

Is pvlib/pvarray.py also eligible for updating? See

pvlib-python/pvlib/pvarray.py

Lines 484 to 486 in cf49dea

	# Eq 9-10
	del0 = (1 - beta_voc * t0) / (50.1 - alpha_sc * t0)
	w0 = np.real(lambertw(np.exp(1/del0 + 1)))

[kandersolar]

LGTM