refactor(model-routing): centralize explicit foreground fallback policy (#6020)

* refactor(model-routing): centralize explicit foreground fallback policy

Make foreground fallback an explicit per-user, availability-only policy shared by streaming Chat, non-stream Chat, and Agent runs.

Preserve strict defaults, owner/model and credential boundaries, pinned Agent routes, and truthful per-round provenance/accounting. Carry provider-reported model identifiers through native streaming adapters, non-stream responses, and caches, and keep legacy default_model_fallbacks as tombstoned raw storage that generic settings APIs and agent tools cannot expose or mutate.

* fix(agent-loop): restore rebase-dropped qwen routing, workspace prompt, and temperature clamp

* fix(model-routing): thread selected endpoint identity, fix cost classification and fallback eligibility

* fix(chat): restore stream helpers and harden run stop lifecycle

* fix(model-routing): let numeric provider codes win over symbolic rate-limit statuses

* fix(agent-loop): apply qwen temperature and notes-tool clamps per fallback candidate

* fix(chat): honor queued stop across resend and reload canonical terminal on EOF

* fix(chat): track stop queue and cleanup ownership by per-send generation

* fix(agent-loop): preserve requested temperature for non-qwen fallback candidates

* fix(chat): reserve send ownership before any await and scope stop to the current send

* fix(chat): clear the previous run identity at send reservation

---------

Co-authored-by: RaresKeY <158580472+RaresKeY@users.noreply.github.com>
Co-authored-by: StressTestor <212606152+StressTestor@users.noreply.github.com>

tests/test_prefs_single_user_no_clobber.py

@@ -6,6 +6,9 @@ every other user's preferences (a realistic ops transition: auth turned off
 on a deployment that previously ran multi-user). It must preserve the other
 users and round-trip the change into the same (first) slot _load_for_user
 reads from.
+
+Foreground fallback keys are the exception: auth-disabled consent is stored
+at the flat root so it can never become consent for the first named owner.
 """
 import json
 
@@ -51,3 +54,58 @@ def test_named_user_save_unaffected(tmp_path, monkeypatch):
     data = json.loads(f.read_text())
     assert data["_users"]["alice"] == {"theme": "light"}
     assert data["_users"]["bob"] == {"theme": "dark"}
+
+
+def test_auth_disabled_fallback_consent_does_not_mutate_first_named_user(
+    tmp_path,
+    monkeypatch,
+):
+    f = tmp_path / "user_prefs.json"
+    f.write_text(json.dumps({"_users": {
+        "alice": {"theme": "light"},
+        "bob": {"theme": "paper"},
+    }}), encoding="utf-8")
+    monkeypatch.setattr(pr, "PREFS_FILE", str(f))
+
+    current = pr._load_for_user(None)
+    current["foreground_fallback_enabled"] = True
+    current["foreground_model_fallbacks"] = [
+        {"endpoint_id": "single-user", "model": "single-model"},
+    ]
+    pr._save_for_user(None, current)
+
+    data = json.loads(f.read_text(encoding="utf-8"))
+    assert data["foreground_fallback_enabled"] is True
+    assert data["foreground_model_fallbacks"][0]["endpoint_id"] == "single-user"
+    assert data["_users"]["alice"] == {"theme": "light"}
+    assert data["_users"]["bob"] == {"theme": "paper"}
+
+
+def test_auth_disabled_save_preserves_named_fallback_consent(tmp_path, monkeypatch):
+    f = tmp_path / "user_prefs.json"
+    alice_fallbacks = [{"endpoint_id": "alice", "model": "alice-model"}]
+    f.write_text(json.dumps({"_users": {
+        "alice": {
+            "theme": "light",
+            "foreground_fallback_enabled": True,
+            "foreground_model_fallbacks": alice_fallbacks,
+        },
+    }}), encoding="utf-8")
+    monkeypatch.setattr(pr, "PREFS_FILE", str(f))
+
+    current = pr._load_for_user(None)
+    assert "foreground_fallback_enabled" not in current
+    assert "foreground_model_fallbacks" not in current
+    current["theme"] = "dark"
+    current["foreground_fallback_enabled"] = False
+    current["foreground_model_fallbacks"] = []
+    pr._save_for_user(None, current)
+
+    data = json.loads(f.read_text(encoding="utf-8"))
+    assert data["foreground_fallback_enabled"] is False
+    assert data["foreground_model_fallbacks"] == []
+    assert data["_users"]["alice"] == {
+        "theme": "dark",
+        "foreground_fallback_enabled": True,
+        "foreground_model_fallbacks": alice_fallbacks,
+    }