HS: runtimeErrors generator patch (+18 tests)

Add `await error(` pattern to generate_eval_only_test — maps
expect(await error("EXPR")).toBe("MSG") to (assert= (eval-hs-error "EXPR") "MSG").
Regenerate behavioral tests; 18 runtimeErrors stubs become real assertions.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/playwright/generate-sx-tests.py

@@ -2318,6 +2318,13 @@ def generate_eval_only_test(test, idx):
     lines = []
     safe_name = sx_name(test['name'])
 
+    # runtimeErrors: expect(await error("EXPR")).toBe("MSG") → eval-hs-error
+    if 'await error(' in body:
+        error_pats = re.findall(r'expect\(await error\("([^"]+)"\)\)\.toBe\("([^"]+)"\)', body)
+        if error_pats:
+            asserts = '\n'.join(f'    (assert= (eval-hs-error "{e}") "{m}")' for e, m in error_pats)
+            return f'  (deftest "{safe_name}"\n    (hs-cleanup!)\n{asserts})'
+
     # Special case: cluster-33 cookie tests. Each test calls a sequence of
     # `_hyperscript("HS")` inside `page.evaluate(()=>{...})`. The runner backs
     # `cookies` with a Proxy over a per-test `__hsCookieStore` map (see