Use Ahem font in /css/css-view-transitions/old-content-is-inline.html for line of text that's not expected to linewrap

[dholbert]

Test List

https://wpt.fyi/results/css/css-view-transitions/old-content-is-inline.html

Rationale

1. The test has a long string of text, rendered in the default font. ("FILLER FILLER INLINE INLINE INLINE INLINE", where "INLINE" is transparent)
2. The reference case has an implicit and unjustified assumption that the text does not line-wrap.
3. In Firefox on the wpt.fyi test runners, with the default font, the text does linewrap; and we fail the test as a result.

Note that if I make the text a bit longer so that it reliably linewraps in all browsers, then it fails in all browsers.

I propose that we adjust the test to use the Ahem font (and also reduce the length of the text so that the Ahem glyphs will fit on a single line).

[dholbert]

Note that if I make the text a bit longer so that it reliably linewraps in all browsers, then it fails in all browsers.

Here's a screenshot to demonstrate this - I edited the testcase and referene case to replace all instances of

INLINE INLINE INLINE INLINE

with

INLINE INLINE INLINE INLINE INLINE INLINE

(6 repetitions rather than 4) so that the text would be long enough to linewrap in Chrome, and here's how the testcase (top) vs. reference case (bottom) looks in Chrome -- notice the bottom part of the testcase ("END STATE") has the green bar in a different position and linewrapped in the testcase, as compared to the reference case:

(Firefox fails in a slightly different way; we don't render the green bar at the bottom of the testcase at all, which is tracked in https://bugzilla.mozilla.org/show_bug.cgi?id=2012351 and tested by other WPTs. But even if that bug were fixed, we'd still fail this particular test because of the test-bug described here; we'd correctly render the green bar at different location in testcase vs. reference case, similar to the Chrome screenshot above. If we edit the test to avoid linewrapping, as I propose doing here, it makes the test more valid, and also happens to avoid that Firefox bug.)

[dholbert]

Got sign-off from Chromium & WebKit representatives in the PR -- thanks folks!