ENH: optimize: support undocumented option full_output for optimize.curve_fit.

[AtsushiSakai]

Reference issue

Fix #6772

What does this implement/fix?

As reported in #6772, optimize.curve_fit has an undocumented option full_output here:

scipy/scipy/optimize/minpack.py

Lines 787 to 788 in 47bb6fe

	# Remove full_output from kwargs, otherwise we're passing it in twice.
	return_full = kwargs.pop('full_output', False)

I think the reason why the option is undocumented is that this option works only when the method is lm, which means using optimize.leastsq.
When one of the other methods is used, it returns OptimizeResult, this is different from the expected full output:

scipy/scipy/optimize/minpack.py

Lines 836 to 837 in 47bb6fe

	if return_full:
	return popt, pcov, infodict, errmsg, ier

This PR enables users to use the option full_output regardless of the used method by creating the expected full output from the OptimzeResult.
Unfortunately, some kinds of output are not included in the OptimizeResult.
So these are ignored, but the full output structure is unified for all methods, the full_output can be officially supported.

[tupui]

Thanks @AtsushiSakai! Some vey minor comments.

[tupui]

LGTM, thanks @AtsushiSakai

[tupui]

@AtsushiSakai can you add a release note in https://github.com/scipy/scipy/wiki/Release-note-entries-for-SciPy-1.9.0? Thanks