Update the symlink creation

[ForestEckhardt]

Attempts to get rid of sorting and make the algorithm more graph like. This will hopefully make the algorithm more robust by removing the significance of ordering of the files.

Resolves #224

Checklist

• I have viewed, signed, and submitted the Contributor License Agreement.
• I have linked issue(s) that this PR should close using keywords or the Github UI (See docs)
• I have added an integration test, if necessary.
• I have reviewed the styleguide for guidance on my code quality.
• I'm happy with the commit history on this PR (I have rebased/squashed as needed).

[ForestEckhardt]

@paketo-buildpacks/tooling-maintainers I only have two question for code review.

1. Should I refactor symlink writing into its own function to remove the duplicate code from the tar and zip archive files?
2. Should I put in some safeguarding against a cyclical symlink situation? And if so what would you suggest for a safeguard strategy?

[ryanmoran]

1. yes. maybe separate the sorting from the symlink creation in the process
2. you can write the loop so that you attempt to make 1 link each time around the outer loop. here is what I was thinking

maxIterations := len(symlinkMap)
for i := 0; i < maxIterations; i++ {
  for path, linkname := range symlinkMap {
    if shouldSkipLink {
      continue
    }

    // make the link
    delete(symlinkMap, path)
    break
  }
}

if len(symlinkMap) > 0 {
  return errors.New("this symlink graph contains a cycle!")
}

[ForestEckhardt]

@ryanmoran Great. I will ping you when I have finished up those changes

[ForestEckhardt]

@ryanmoran I added to logic to both remove the potential for deadlock as well as removing the label. I didn't quite grock what you wanted in terms of extracting the functionality so I left it in place for the time being.

[ryanmoran]

@ForestEckhardt here is what I was thinking: e298983

[ForestEckhardt]

@ryanmoran Honestly that refactor works for me! If that is all good on your end then we can just merge.