Mirror the rubocop-rails warning silence workaround

[mjankowski]

This is an attempt at resolving #72

As noted in that issue, I only vaguely understand the rubocop rule loading and validating process -- and then both rubocop-rails and standard-rails do some amount of overriding/modifying that process. Anyway, this is a first pass and maybe it's fine or maybe we can find something more elegant.

Concerns here:

• This change is inspired by - https://github.com/rubocop/rubocop-rails/blob/v2.33.3/lib/rubocop/rails/plugin.rb#L27-L32 - which is doing the same sort of warning silencing from within rubocop rails. I don't love the idea of borrowing something which calls itself a dirty hack.
• I'm not actually sure how to produce these warnings from the standard rails test suite itself (I only see it from the full loading process initiated by a separate app), so I'm not sure what test to even add here to avoid regression and/or let us make this more elegant in the future
• I don't really understand how we DONT see a warning for TargetRailsVersion (the one worked around in rubocop-rails), since (I think?!) we are overriding the load process such that the rubocop-rails #rules method never gets called, and so, one would think, whatever issue that one is working around would need to be re-worked-around?

Benefits:

• If rubocop-rails is doing this internally, maybe it's good enough? (and presumably if/when some cleaner solution emerges we could borrow it?)
• This does indeed silence that output from a local app which otherwise sees the issue, when I update to point at local repo of standard rails with this change

Random side adventure during this journey ... in the other part of this diff, where I added Lint/UselessAccessModifier to the existing list to disable for similar reasons, there's some small inconsistency with the comment there and the rules and the history of that whole thing, I think?

• The comment says standard does not enable them to begin with, which is true for Style/InvertibleUnlessCondition, but not true for Lint/SafeNavigationChain which as of https://github.com/standardrb/standard-rails/pull/22/files#diff-da3213a24b350fa386ff3bab0d6b239112408c3fe4f986cbd9d5236f7e0b61e4R31-R40 is enabled
• The commit which added this "without extended" - acccb02 - removes Lint/RedundantSafeNavigation and Style/InvertibleUnlessCondition from the base config, but then adds Style/InvertibleUnlessCondition and Lint/SafeNavigationChain to this reject line ... should that latter addition have actually been Lint/RedundantSafeNavigation (to match what was removed), and/or should they all be there?
• At minimum maybe that comment needs an update.

Thoughts on all topics welcome.

[mjankowski]

Ah -- funny timing, basically dupe of #97

Will leave open until that's reviewed.

[mjankowski]

Added one of the tests from linked PR, which helped expose the "why wouldnt we have to ignore target version as well?" issue.

Updated to include that test, and a change needed to also remove target version thing.

[jasonkarns]

Thank you for this! Just merged #97, though, so closing this one.

[mjankowski]

Thanks.

Any thoughts on my last note? Should I open separate issue for that.

[jasonkarns]

@mjankowski yes, a separate issue would be welcome!! apologies if I jumped too quickly to close

[mjankowski]

No problem, I would have closed this when the other merged anyway.

Will attempt to turn the relevant bits into a coherent issue.