Fix rendering issue for links with special characters

[d-ronnqvist]

Bug/issue #, if applicable: rdar://85531439

Summary

Fix an issue where symbol links or documentation links containing special characters resolved successfully but didn't render as expected on the page.

There are two parts to this fix:

• Allow any previously resolved reference to be looked up without resolving it again. I do this because MarkupReferenceResolver replaces the links original destination with the absolute string of the resolves reference but the resolved reference replaces non-path characters with "-" resulting in potential ambiguity when re-resolving the link.
• Preserve the isAutolink attribute of Link values when updating their destination with a new resolved destination

Dependencies

None.

Testing

• Use one or more special characters—for example π or 😃—in an element that can be linked to. This can be either in a symbol name, article file name, or heading.
  See for example this unit test.
• Write a symbol link to the symbol, article, or heading containing the special character.
  Expected: The link should resolve and render with its title as the link text
• Also curate the symbol, article, or heading containing the special character.
  Expected: The link should resolve and render with its title as the link text

Checklist

Make sure you check off the following items. If they cannot be completed, provide a reason.

• Added tests
• Ran the ./bin/test script and it succeeded
• Updated documentation if necessary

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

Despite adding a lookup of all resolved references I've not seen any regression in peak memory usage in my local testing

┌──────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ Metric                                   │ Change          │ main                 │ current              │
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ Duration for 'bundle-registration'       │ no change¹      │ 5.635 sec            │ 5.644 sec            │
│ Duration for 'convert-total-time'        │ no change²      │ 9.626 sec            │ 9.613 sec            │
│ Duration for 'documentation-processing'  │ no change³      │ 3.758 sec            │ 3.738 sec            │
│ Duration for 'finalize-navigation-index' │ no change⁴      │ 0.117 sec            │ 0.116 sec            │
│ Peak memory footprint                    │ no change⁵      │ 1,014.9 MB           │ 1,008.7 MB           │
│ Data subdirectory size                   │ no change       │ 280.4 MB             │ 280.4 MB             │
│ Index subdirectory size                  │ no change       │ 2.5 MB               │ 2.5 MB               │
│ Total DocC archive size                  │ no change       │ 371.7 MB             │ 371.7 MB             │
│ Topic Anchor Checksum                    │ no change       │ a8c40d797627683c5d7c │ a8c40d797627683c5d7c │
│ Topic Graph Checksum                     │ no change       │ da82bd7f35eae4bb56e8 │ da82bd7f35eae4bb56e8 │
└──────────────────────────────────────────────────────────────────────────────────────────────────────────┘

 1: No statistically significant difference.
    The values are similar enough that the most probable explanation is that they're random samples from the same data set.
    t-statistic : -0.387          degrees of freedom : 28       95% confidence critical value : +2.042

 2: No statistically significant difference.
    The values are similar enough that the most probable explanation is that they're random samples from the same data set.
    t-statistic : +0.494          degrees of freedom : 28       95% confidence critical value : +2.042

 3: No statistically significant difference.
    The values are similar enough that the most probable explanation is that they're random samples from the same data set.
    t-statistic : +1.539          degrees of freedom : 28       95% confidence critical value : +2.042

 4: No statistically significant difference.
    The values are similar enough that the most probable explanation is that they're random samples from the same data set.
    t-statistic : +1.669          degrees of freedom : 28       95% confidence critical value : +2.042

 5: No statistically significant difference.
    The values are similar enough that the most probable explanation is that they're random samples from the same data set.
    t-statistic : +1.792          degrees of freedom : 28       95% confidence critical value : +2.042

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

Hm.. for some reason the CI is failing with errors that I can't reproduce locally with either Swift 5.8 or Swift 5.9

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

@swift-ci please test

[d-ronnqvist]

After the latest changes I still see no measurable performance differences compared to main.

[daniel-grumberg]

I don't remember what version of swift we aim to support, would some StringProtocol be acceptable here?

[d-ronnqvist]

Not yet, we support Swift 5.5 and the some keyword was added in Swift 5.7

[d-ronnqvist]

@swift-ci please test