test(materialize): pin the atBlank patch's equivalence under V8 too (TASK-2198 U2, codex r1)

Codex r1 claimed upstream's /(^|\n)$/ is true for a trailing CR in JS. It is
not (checked in V8: false for "\r" and "a\u2028", as the patch returns),
but the only equivalence test ran under goja and the patch header cited a
V8-side test that did not exist. This adds it, with the line-terminator edge
cases, and it fails when the patch checks CR instead of LF.

Claude-Session: https://claude.ai/code/session_01H63zaWhkb5pAafx2WGjrVC

web/src/lib/collab/materializer/atBlank.test.ts

@@ -0,0 +1,30 @@
+import { describe, it, expect } from 'vitest';
+import { upstreamAtBlank, patchedAtBlank } from './atBlank';
+
+// The vendored atBlank patch must agree with upstream prosemirror-markdown in
+// the engine a browser tab runs, not only in goja (internal/materialize has
+// the goja leg). This runs under V8. The line-terminator cases are the ones
+// where ECMAScript `$` could plausibly differ from an ends-with-LF check: it
+// does not, without the `m` flag, and this pins that.
+describe('atBlank patch equivalence under V8 (TASK-2198)', () => {
+	const call = (fn: typeof upstreamAtBlank, out: string) => fn.call({ out });
+
+	it('agrees on line-terminator edge cases', () => {
+		const cases = ['', '\n', '\r', '\r\n', '\n\r', '
', '
', 'a', 'a\n', 'a\r', 'a\r\n', 'a
', 'a\n\n', ' \n', '\n '];
+		for (const out of cases) {
+			expect(call(patchedAtBlank, out), JSON.stringify(out)).toBe(call(upstreamAtBlank, out));
+		}
+	});
+
+	it('agrees on generated strings', () => {
+		const alphabet = ['a', ' ', '\n', '\r', '\t', '
', '#', '`'];
+		let seed = 3252;
+		const rnd = () => (seed = (seed * 1103515245 + 12345) & 0x7fffffff) / 0x7fffffff;
+		for (let i = 0; i < 5000; i++) {
+			const len = Math.floor(rnd() * 12);
+			let out = '';
+			for (let j = 0; j < len; j++) out += alphabet[Math.floor(rnd() * alphabet.length)];
+			expect(call(patchedAtBlank, out), JSON.stringify(out)).toBe(call(upstreamAtBlank, out));
+		}
+	});
+});