docs: the last two tests that credit the sidecar for an iOS fact

Same correction as the alias table they cover: accessibilityText comes from the
companion.

internal/driver/chrome/selector_parity_test.go

@@ -119,7 +119,7 @@ func TestSelectors_ResolveTheSameElementsAsTheWebRuntime(t *testing.T) {
 			// Every name for the accessible label has to name one element. Four
 			// of the six reached it on one host only: label and
 			// accessibilityLabel aliased onto accessibilityText alone, which
-			// the ios sidecar writes and this dump does not, and alias
+			// the ios companion writes and this dump does not, and alias
 			// expansion is one level; ariaLabel and contentDescription aliased
 			// onto nothing. accessibilityText was the mirror image, resolving
 			// against the dump and reaching no DOM attribute of that name.

internal/hierarchy/hierarchy_test.go

@@ -695,7 +695,7 @@ func TestAccessibilityLabelAliasesReachContentDesc(t *testing.T) {
 	}
 }
 
-// The iOS sidecar writes the same fact under accessibilityText, which the two
+// The iOS companion writes the same fact under accessibilityText, which the two
 // names already reached and have to keep reaching.
 func TestAccessibilityLabelAliasesStillReachAccessibilityText(t *testing.T) {
 	tree, _ := Parse(iosAttrDump)