Reduce appearance tests in Web Compat to 10 tests picked at random

[foolip]

These were picked with the following command:

cat original-list-of-tests | sort -R | head -n10 | sort

[foolip]

Here's what the focus area would look like if we merge this:
https://wpt.fyi/results/?label=master&label=experimental&view=interop&q=label%3Ainterop-2025-webcompat%20and%20%28%21compute-kind-widget-generated%20or%20kind-of-widget-fallback-button-border-block-end-width-001.html%20or%20kind-of-widget-fallback-button-border-image-slice-001.html%20or%20kind-of-widget-fallback-button-border-image-source-001.html%20or%20kind-of-widget-fallback-input-reset-border-right-color-001.html%20or%20kind-of-widget-fallback-input-submit-border-block-end-color-001.html%20or%20kind-of-widget-fallback-input-submit-border-image-repeat-001.html%20or%20kind-of-widget-fallback-input-text-border-bottom-right-radius-001.html%20or%20kind-of-widget-fallback-input-text-border-start-start-radius-001.html%20or%20kind-of-widget-fallback-meter-background-attachment-001.html%20or%20kind-of-widget-fallback-radio-input-border-bottom-left-radius-001.html%29

[gsnedders]

This would cause us to end up with these results.

That results in, for compute-kind-widget-generated specifically:

	Old	New	Diff
Chrome	100%	100%	0%
Edge	100%	100%	0%
Firefox	75.2%	60%	-15.2%
Safari	89.3%	80%	-9.3%

[Edit: hah, beaten by @foolip by virtue of trying to write a table summarising the change!]

[jgraham]

I think this is fine as a placeholder change, but we should work out a way to get more coverage in a reasonable number of tests.

[gsnedders]

There's 45 other tests, so this takes us from 94.6% of the Web Compat focus area to 18.2%.

Given the current scores of the whole directory, any subset of 10 tests here will alter the Firefox score for the directory by at least 5%. We could go up to 20 tests, which would be 30.8% of the focus area, and be able to better approximate the current scores. That said, I'm not trying to assert we change the current plan to move to 10 tests.

Let's cross-reference this to web-platform-tests/rfcs#167 given subtests might help. It's also potentially related to broader questions about balancing scoring across features within a focus area for Interop, and this is not the first time this has come up.

[gsnedders]

FWIW, per-element:

Element	Chrome	Edge	Firefox	Safari
select-dropdown-box	100.0%	100.0%	22.7%	63.6%
select-menulist-button	100.0%	100.0%	22.7%	100.0%
progress	100.0%	100.0%	59.1%	100.0%
button	100.0%	100.0%	77.3%	72.7%
input-button	100.0%	100.0%	77.3%	72.7%
input-reset	100.0%	100.0%	77.3%	72.7%
input-submit	100.0%	100.0%	77.3%	72.7%
color-input	100.0%	100.0%	77.3%	72.7%
input-search	100.0%	100.0%	77.3%	81.8%
meter	100.0%	100.0%	77.3%	100.0%
textarea	100.0%	100.0%	77.3%	100.0%
input-text	100.0%	100.0%	77.3%	100.0%
select-listbox	100.0%	100.0%	77.3%	100.0%
input-search-text	100.0%	100.0%	77.3%	100.0%
link	100.0%	100.0%	100.0%	100.0%
range	100.0%	100.0%	100.0%	100.0%
checkbox-input	100.0%	100.0%	100.0%	100.0%
radio-input	100.0%	100.0%	100.0%	100.0%

And per-property:

Property	Chrome	Edge	Firefox	Safari
background-attachment	100.0%	100.0%	33.3%	100.0%
background-clip	100.0%	100.0%	33.3%	100.0%
background-origin	100.0%	100.0%	33.3%	100.0%
background-position	100.0%	100.0%	33.3%	100.0%
background-size	100.0%	100.0%	33.3%	100.0%
border-image-outset	100.0%	100.0%	33.3%	100.0%
border-image-repeat	100.0%	100.0%	33.3%	100.0%
border-image-slice	100.0%	100.0%	33.3%	100.0%
border-image-source	100.0%	100.0%	33.3%	100.0%
border-image-width	100.0%	100.0%	33.3%	100.0%
border-block-end-width	100.0%	100.0%	83.3%	66.7%
border-block-start-width	100.0%	100.0%	83.3%	66.7%
border-bottom-width	100.0%	100.0%	83.3%	66.7%
border-top-width	100.0%	100.0%	83.3%	66.7%
border-block-end-color	100.0%	100.0%	88.9%	66.7%
border-block-end-style	100.0%	100.0%	88.9%	66.7%
border-block-start-color	100.0%	100.0%	88.9%	66.7%
border-block-start-style	100.0%	100.0%	88.9%	66.7%
border-bottom-color	100.0%	100.0%	88.9%	66.7%
border-bottom-style	100.0%	100.0%	88.9%	66.7%
border-top-color	100.0%	100.0%	88.9%	66.7%
border-top-style	100.0%	100.0%	88.9%	66.7%
border-inline-end-width	100.0%	100.0%	83.3%	94.4%
border-inline-start-width	100.0%	100.0%	83.3%	94.4%
border-left-width	100.0%	100.0%	83.3%	94.4%
border-right-width	100.0%	100.0%	83.3%	94.4%
border-inline-end-color	100.0%	100.0%	88.9%	94.4%
border-inline-end-style	100.0%	100.0%	88.9%	94.4%
border-inline-start-color	100.0%	100.0%	88.9%	94.4%
border-inline-start-style	100.0%	100.0%	88.9%	94.4%
border-left-color	100.0%	100.0%	88.9%	94.4%
border-left-style	100.0%	100.0%	88.9%	94.4%
border-right-color	100.0%	100.0%	88.9%	94.4%
border-right-style	100.0%	100.0%	88.9%	94.4%
background-color	100.0%	100.0%	88.9%	100.0%
background-image	100.0%	100.0%	88.9%	100.0%
border-bottom-left-radius	100.0%	100.0%	88.9%	100.0%
border-bottom-right-radius	100.0%	100.0%	88.9%	100.0%
border-end-end-radius	100.0%	100.0%	88.9%	100.0%
border-end-start-radius	100.0%	100.0%	88.9%	100.0%
border-start-end-radius	100.0%	100.0%	88.9%	100.0%
border-start-start-radius	100.0%	100.0%	88.9%	100.0%
border-top-left-radius	100.0%	100.0%	88.9%	100.0%
border-top-right-radius	100.0%	100.0%	88.9%	100.0%

This doesn't make it super obvious what the one test for Safari should be, nor what the two/three tests for Firefox should be.