FIX - add alpha to SLOPE penalty

[jolars]

Context of the PR

See #315

Contributions of the PR

Add alpha to make repr() work on objects created with the SLOPE penalty. Closes #315.

Checks before merging PR

• added documentation for any new feature
• added unit tests
• edited the what's new (if applicable)

[Badr-MOUFAD]

LGTM, thx @jolars

[mathurinm]

Sorry for the delay, my only concern is that this introduces redundancy
but for a fixed decay ratio this may make parameter tuning easier (only one param, alpha, to tune).

@jolars is it handled the same way in other packages ?

[jolars]

Actually I don't really have an opinion on how to handle this. My only interest in this is really to fix the bug and prevent the benchopt tests from erroring out 😄 . So I'm happy with any solution, really. That being said, I think it's worthwhile to have alpha be a parameter since it is almost almost this parameter along with whatever other variables (such as q) you might use to parameterize the penalty sequence. I do, however, think that the use of both alpha and alphas to mean essentially different things is a little confusing, but this way at least you don't break anything.

[jolars]

I forgot to answer your question. Yes, at least the packages that I have made always have both an argument for the scaling of the penalty sequence and the penalty sequence itself. Yes it's redundant, but also practical.