docs(r2): correct the provenance of the rewritten keyboard-flag test

The gallery's POST-303-reload was the no-JS design working, and it still
is (the INV-4 golden pins it). The defect was the full-size ejection. The
test's docstring and the contract's assertions table now say so (booth-dev
review).

tests/test_embed_browser.py

@@ -595,11 +595,13 @@ def test_the_keyboard_flag_actually_submits(browser, live):
     Asserted end to end: press f, and the flag must come back from the server.
 
     R2 C3 (docs/contracts/r2_flow.contract.md, "Assertions that change"): this
-    used to expect a NAVIGATION — the flag form POSTed and the page reloaded.
-    That reload is the defect R2 removes. The claim that matters is kept and
-    tightened: the flag must come back from the SERVER (the swapped tile is
-    server-rendered), and a marker set on the window before the keypress must
-    survive, which a reload would wipe."""
+    used to expect a NAVIGATION — the flag form POSTed, 303'd and reloaded. That
+    was the no-JS design working, not a defect, and it still is with scripts
+    off (tests/golden/r2_mark_303.json replays those responses byte for byte).
+    What changed is that WITH JS ON the flag now applies in place. The claim
+    that matters is kept and tightened: the flag must come back from the SERVER
+    (the swapped tile is server-rendered), and a marker set on the window before
+    the keypress must survive, which a reload would wipe."""
     base, root = live
     _gallery(root)
     page = browser.new_page()