object_length_linter should detect generics

[AshesITR]

As the object_name_linter already does, class suffixes for S3 generics shouldn't be part of the linted symbol.
That is, object_length_linter(30) (the default) shouldn't lint this example:

my_generic_function <- function(x, ...) {
  UseMethod("my_generic_function")
}

my_generic_function.my_cool_class <- function(x, ...) {
  cat("How cool and generic!\n")
}

[stufield]

Yes! I've been bitten by this many times and have # nolint scattered throughout my generics definition files.
Looking forward to this.

[AshesITR]

For now, I've even disabled the linter alltogether in the .lintr config because it didn't have any true positives in my project.

[MichaelChirico]

Should this be an option?

[AshesITR]

I'd be happy to make it optional. What should the default be? The old behaviour?

[MichaelChirico]

I think the new behavior is a happier default -- often it's out of your control if you're defining a method for someone else's class.

WDYT about a third option -- enforce the length on two parts (class name and generic name)? not sure it's worth the implementation & maintenance effort but it does make some sense

[AshesITR]

We could combine both by specifying two maximum lengths?