minor l-curve comments

[asiaszmek]

1. estimate='CSD'is a parameter of L_curve, it is not used anywhere.
2. L_curve does not return anything despite the statement in the docstrings.
3. decimal=3 in prints, skcsd uses source widths of the order of tens of microns.
4. Rs is converted to an array, lambdas isn't, it would be quite useful, if it was.

[ccluri]

Is there any analysis of L-Curve predictions with time?
An example of this would be nice.

[ccluri]

@asiaszmek are these addressed in #70 ? Can we close this?

[ccluri]

Bump @asiaszmek