wpt: decode non-UTF-8 test files honouring BOM and declared encoding

[nicoburns]

Summary

The runner crashed with stream did not contain valid UTF-8 on 11 css tests encoded as UTF-16 or a legacy encoding (css/CSS2/i18n/syndata/character-encoding-*.xht, css/css-syntax/charset/page-utf16-*.html), because process_test_file used fs::read_to_string(...).unwrap().

Test files are now read as bytes and decoded via decode_file_bytes:

1. BOM (via encoding_rs::Encoding::for_bom) wins;
2. otherwise valid UTF-8 is used as-is;
3. otherwise the first 1024 bytes are scanned for a declared encoding (encoding=/charset= in an XML declaration, <meta charset>, or @charset), with a windows-1252 fallback. A declared UTF-16 label maps to UTF-8, per the HTML spec's "get an encoding" algorithm (an ASCII-readable declaration means the content isn't UTF-16).

WPT crash counts for css/CSS2/i18n/syndata + css/css-syntax/charset: 11 CRASH before, 0 after (all now run/skip normally).

Note: 4 tests flip PASS→FAIL with this change (font-family-invalid-characters-001/003.xht, text-emphasis-style-none-001.xht, text-transform-none-001.xht). These files have a UTF-8 BOM which was previously left in the parser input, corrupting parsing so their stylesheets never applied and they trivially matched their (unstyled) refs. With the BOM correctly stripped they actually exercise their styles and expose a separate engine issue: text consisting only of digits is not rendered when the font-family stack contains only unknown families (e.g. font-family: test) — letters fall back correctly but digits disappear.

Link to Devin session: https://dioxus.staging.devinenterprise.com/sessions/f1a0d6696eaa4076a48f366a24bfd26f
Requested by: @nicoburns

WPT results

0 newly passing, 4 newly failing (net -4), 11 other status changes.

Full diff (15 changed tests)

- Pass => Fail css/CSS2/fonts/font-family-invalid-characters-001.xht
- Pass => Fail css/CSS2/fonts/font-family-invalid-characters-003.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-003.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-006.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-009.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-013.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-014.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-015.xht
! Crash => Skip css/CSS2/i18n/syndata/character-encoding-016.xht
! Crash => Skip css/css-syntax/charset/page-utf16-css-bomless-utf16.html
! Crash => Skip css/css-syntax/charset/page-utf16-css-bomless-utf16be.html
! Crash => Skip css/css-syntax/charset/page-utf16-css-no-decl-ascii-only.html
! Crash => Skip css/css-syntax/charset/page-utf16-css-no-decl.html
- Pass => Fail css/css-text-decor/text-emphasis-style-none-001.xht
- Pass => Fail css/css-text/text-transform/text-transform-none-001.xht

_{Generated by the WPT workflow.}

[staging-devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR that start with 'DevinAI' or '@devin'.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring