Removes glob import of SnapshotKind

[brooksprumo]

Problem

Well... today I learned about glob imports of enums and how they interact with match statements. Basically, glob imports of an enum that are used with a match can be a footgun. Refer to this playground:

https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=9f0ce5f074f7b4a71f487ae8c6a1902d

Summary of Changes

Removes glob import of SnapshotKind and AccountsPackageKind.

[codecov]

Codecov Report

Merging #34333 (c700439) into master (5c00a33) will increase coverage by 0.0%.
Report is 2 commits behind head on master.
The diff coverage is 100.0%.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #34333   +/-   ##
=======================================
  Coverage    81.9%    81.9%           
=======================================
  Files         819      819           
  Lines      220422   220422           
=======================================
+ Hits       180652   180707   +55     
+ Misses      39770    39715   -55     

[apfitzge]

Not opposed to the changes but think that naming conventions and the warnings from violating those conventions are generally sufficient to protect against this kind of mistake.

If I read your example correctly, it's not so much to do with global imports but just that 'C' is acting as the catch all case. Explicit namespacing guarantees we don't make that mistake, but it would still generate a warning since 'C' is both unused and a non snake_case variable name.

[brooksprumo]

If I read your example correctly [..]

Yep, you did!

Not opposed to the changes but think that naming conventions and the warnings from violating those conventions are generally sufficient to protect against this kind of mistake.

I'm mostly thinking about preventing incorrect refactorings. In the example, if I change the match to only resolve the warnings, I could have something like this:

        match self {
            A => write!(f, "A"),
            B => write!(f, "B"),
            _c => write!(f, "C"),
        }

Which is still broken. Hopefully this code change would never make it past code review. I feel safer to make this entire class of problems impossible, by removing the glob import.

Is there another implementation and/or tweak that you would be in favor of?