effective mass fitting improve

[zhubonan]

Improve the robustness of effective mass fitting algorithm.
CLI and notebook examples are added.

[kavanase]

Thanks for getting to this @zhubonan!

Looks mostly good to me. Some small things:

• Could we add some docstrings, specifically to get_effective_masses (used in the example) and get_band_extrema, EffectiveMass.
• The CLI eff mass output ends with:

Unfolded band structure can be ambiguous, please cross-check with the spectral function plot.

This isn't very clear. Can we update this? Cross-check what with the spectral function plot? That the eff mass (or extrema) values make sense from eyeballing?

• The eff mass example ends with this:

To obtain reliable effective masses, one should generate finely spaced kpoint paths that includes only the vicinity of the CBM/VBM.

This is a bit ominous on its own without any advice to ensure this. Can we add something here about what the user should check? (And I guess if it's not likely converged, then the advice is the user should recalculate the band structure with denser kpoints?)

[kavanase]

This looks great, thanks for the updates @zhubonan!

[kavanase]

Is this ready to merge @zhubonan?
I guess it solves #5 so can also close that. Might be worth adding whatever docstrings/notes you wanted to add for #38 now as well before merging? 😃

[zhubonan]

Yeah, will see if I can sort out #38 here as well.

[kavanase]

Great, looks good to me! Let's go ahead and merge this so? 😃