Bugzilla

I can reproduce the failure locally. With `layout.css.at-scope.enabled` set to `true`, this test has 10 subtests pass and 1 fail, with the failure being:
```
Fail	@scope { & .b { } } and :where(:scope) .b	assert_equals: unscoped + scoped expected "2" but got "1"
```

The upstream PR has this note to explain itself:
```
The '&' selector, when it has no parent selector to draw from,
has a specificity of zero [1].

Blink already does the right thing. This CL just adds a new WPT
to enforce that behavior, and adjusts an existing tests which
expected specificity to be added.
```

dshin, could you take a look as part of your ongoing @scope work?

I can reproduce the failure locally. With `layout.css.at-scope.enabled` set to `true`, this test has 10 subtests pass and 1 fail, with the failure being:
```
Fail	@scope { & .b { } } and :where(:scope) .b

assert_equals: unscoped + scoped expected "2" but got "1"
```

The upstream PR has this note to explain itself:
```
The '&' selector, when it has no parent selector to draw from,
has a specificity of zero [1].

Blink already does the right thing. This CL just adds a new WPT
to enforce that behavior, and adjusts an existing tests which
expected specificity to be added.
```

dshin, could you take a look as part of your ongoing @scope work?

I can reproduce the failure locally. With `layout.css.at-scope.enabled` set to `true`, this test has 10 subtests pass and 1 fail, with the failure reported like so:
```
Fail	@scope { & .b { } } and :where(:scope) .b

assert_equals: unscoped + scoped expected "2" but got "1"
```

The upstream PR has this note to explain itself:
```
The '&' selector, when it has no parent selector to draw from,
has a specificity of zero [1].

Blink already does the right thing. This CL just adds a new WPT
to enforce that behavior, and adjusts an existing tests which
expected specificity to be added.
```

dshin, could you take a look as part of your ongoing @scope work?

I can reproduce the failure locally. With `layout.css.at-scope.enabled` set to `true`, this test ([hosted on wpt.fyi](https://wpt.fyi/results/css/css-cascade/scope-specificity.html)) shows us passing 10 subtests and failing 1, with the failure reported like so:
```
Fail	@scope { & .b { } } and :where(:scope) .b

assert_equals: unscoped + scoped expected "2" but got "1"
```

The upstream PR has this note to explain itself:
```
The '&' selector, when it has no parent selector to draw from,
has a specificity of zero [1].

Blink already does the right thing. This CL just adds a new WPT
to enforce that behavior, and adjusts an existing tests which
expected specificity to be added.
```

dshin, could you take a look as part of your ongoing @scope work?