Add ignore-signatures option to [SIMILARITIES] section

[jnsnow]

Is your feature request related to a problem?

Please describe A clear and concise description of what the problem is.

When comparing code that uses multiple lines for function signatures, it's often the case that classes that implement or extend an interface will have several repeated lines for that interface signature.

e.g. consider a class implementing the __exit__ interface for a context manager:

class ArbitraryExample:
    def __exit__(self,
                 exc_type: Optional[Type[BaseException]],
                 exc_val: Optional[BaseException],
                 exc_tb: Optional[TracebackType]) -> None:
        pass

We're at four lines. If two modules in the project use a context manager, it will trigger the similarity checker.

Describe the solution you'd like

A clear and concise description of what you want to happen.

Implement a 'ignore-signature' option for the [SIMILARITIES] section, and have it default to yes.

(In same cases I imagine it might be a legitimate concern that you have many functions with extremely similar interfaces. In the case of special dunder methods at least, we are generally targeting an interface and have no control over the factoring. An ignore option like this might be best, since it seems hard to accurately determine which cases of repetition are fixable and which are not.)

[adityagupta1089]

Let me know if I can help here.

[Pierre-Sassoulas]

Hey @adityagupta1089 , no one is assigned to the issue so if you want to handle it I can assign it to you :)

[adityagupta1089]

@Pierre-Sassoulas cool, I'll probably need some background or otherwise I can go through the code. Do you have some proposal for a solution or should I propose one?

[Pierre-Sassoulas]

Nice, I assigned you to the issue 😄 There is already an option for the similarity checker for imports, this one will probably be pretty close to it. You could check that here: https://github.com/PyCQA/pylint/blob/master/pylint/checkers/similar.py#L53 .

[adityagupta1089]

@Pierre-Sassoulas I think the PR was merged but did not trigger closure of issue. (I mentioned closes #3619 in description)

[Pierre-Sassoulas]

Thanks for notifying me, cleaning these periodically after the fact is a full day of work :D

[jnsnow]

For my own record, this was addressed in PR #4474

(Thank you for implementing it!)