fix: escape non-ASCII characters as \uNNNN in string literals

[Brooooooklyn]

Match TypeScript's emitter behavior by escaping all non-ASCII characters (code point > 0x7E) as \uNNNN sequences. Characters above the BMP use UTF-16 surrogate pairs. Uses push_str with hex table lookup instead of fmt::Write.

---

Note

Medium Risk
Changes emitted JavaScript string literal contents (non-ASCII now always becomes \u escapes), which can affect snapshot tests, output diffs, and any consumers expecting literal UTF-8 in generated code; logic is localized but touches core emission paths.

Overview
Updates escape_string in the Angular compiler JS emitter to always escape non-ASCII characters (and ASCII control chars) as \uNNNN, emitting UTF-16 surrogate pairs for code points above the BMP to match TypeScript’s string-literal printer.

Adds a small helper (push_unicode_escape) to build escapes without formatting, expands unit coverage around surrogate pairs and non-ASCII cases, and updates integration assertions so compiled templates with HTML entities (e.g. ×,  ) now produce escaped sequences instead of literal Unicode characters.

^{Written by Cursor Bugbot for commit 8d880a6. This will update automatically on new commits. Configure here.}