Only shape point placement symbols

[malwoodsantoro]

Launch Checklist

Fixes #8575

• briefly describe the changes in this PR
  • This PR fixes a bug in which newline characters were being ignored if text-max-width was set to zero. This was caused by a faulty check in the shaping code which equated "no max width" with "no shaping should occur". This PR fixes the issue and corrects the check by explicitly checking for line and line-center symbol placement and not shaping those symbols (e.g. only shape text when the symbol is placed as a point).
• write tests for all new functionality
• document any changes to public APIs
• post benchmark scores
• manually test the debug page
• tagged @mapbox/studio and/or @mapbox/maps-design if this PR includes style spec changes
• tagged @mapbox/gl-native if this PR includes shader changes or needs a native port
  • this will need a port to native cc: @mapbox/gl-native

Changelog
Fix bug in which explicit newline characters were ignored when text-max-width was set to zero

[ryanhamley]

This is working really well now. The only failing test is a render test text-max-width/unlimited which I believe to be an incorrect test. The test uses text-max-width: 0 and equates it with "unlimited" width, which is incorrect. My recommendation is to convert this to an explicit test of text-max-width: 0 with the expectation that there are line breaks on every possible whitespace character.

[ryanhamley]

#3725 is the PR that introduced the if (!maxWidth) check that is being changed here. I don't really understand why Chris went this particular direction. Making this change seems to be more consistent with the expected behavior and removes a confusing semantic equivalence between 0 max width and effectively infinite max width. I think this change is ok, but I just wanted to note the original PR in case.

[asheemmamoowala]

The only failing test is a render test text-max-width/unlimited which I believe to be an incorrect test.
Making this change seems to be more consistent with the expected behavior and removes a confusing semantic equivalence between 0 max width and effectively infinite max width. I think this change is ok

@malwoodsantoro @ryanhamley It is unclear how this change affects existing styles that may be depending on this behavior. Maybe someone from @mapbox/map-design-team can comment on whether this change will require style changes for common use cases.

[ryanhamley]

@mapbox/map-design-team Does anyone have feedback on this change?

[mourner]

Looks good to me, pending benchmark results to make sure we don't regress on performance.

[samanpwbb]

Change makes sense and looks good 👍

[ryanhamley]

Benchmarks look good

[ryanhamley]

Visual Changes

This style with text-max-width: 0 and an explicit newline character in the text:

Produces this output on master. Note that the newline is ignored.

With this PR, this is the output. Since the text max width is 0, we break on every whitespace character.