Add a baseline capability to Periphery

[mildm8nnered]

WIP.

Addresses #710

This is the equivalent of realm/SwiftLint#5475

This adds two new arguments for the scan command: --write-baseline <baseline> to write a baseline file, and --baseline <baseline> to specify a baseline to use while scanning,

Should be functional, but periphery does not give consistent results for me since 2.18.0, which is problematic - see #731

Straight port from the SwiftLint version - style is considerably different to typical periphery code style
Currently no unit tests
No way to view or compare baselines (potentially a nice to have).

[aiKrice]

@mildm8nnered How could I help you to test ? Please provide a how to in the PR description. I would like to help you. I followed you on Swiftlint already :D

[mildm8nnered]

@mildm8nnered How could I help you to test ? Please provide a how to in the PR description. I would like to help you. I followed you on Swiftlint already :D

Hi @aiKrice - I'll add some instructions shortly. Would love to get some feedback.

[mildm8nnered]

@mildm8nnered How could I help you to test ? Please provide a how to in the PR description. I would like to help you. I followed you on Swiftlint already :D

Hi @aiKrice - I'll add some instructions shortly. Would love to get some feedback.

I've updated the description above ...

[ileitch]

Thanks for working on this @mildm8nnered. Have you considered only saving the USR in the baseline? The baseline would still behave the same after some code movement while being much simpler to implement.

[mildm8nnered]

Thanks for working on this @mildm8nnered. Have you considered only saving the USR in the baseline? The baseline would still behave the same after some code movement while being much simpler to implement.

Hi @ileitch - I have not really played with it that much, as I was focussed on getting the SwiftLint version over the line.

If there a way to store less, that would be great. I'm travelling right now, but will be back home later this week, so will pick this back up again then.

[mildm8nnered]

So #731 is resolved for me, and I fixed some issues where I was incorrectly storing the hash values.

I'm still seeing some weirdness and inconsistency in the baseline output, but it looks like something else - violations look almost like they should match between runs, but the path storage and locations are different. Not sure if the problem is in the new code or not, but I'll dig into it.

[ileitch]

@mildm8nnered The complexity of this approach makes me nervous, and I'm worried we might miss some edge cases, or regress easily. I think the USR alone should be enough, at least to begin with. Here's a comparative impletion that just records the USRs: #751

[ileitch]

Implememted in #751. Thanks for guiding this effort, @mildm8nnered!