studio: fix tests turning main CI red/flaky (kill-process, install overrides, UI re-login) (#6419)

* studio: set _stats_logger in kill-process test backend

#6377 added a self._stats_logger cleanup step to _kill_process's finally block.
test_kill_process_records_timestamp_on_actual_kill (added in #6400) builds the
backend via __new__, which bypasses __init__ where _stats_logger is set, so once
both landed on main the test raised AttributeError: 'LlamaCppBackend' object has
no attribute '_stats_logger'. Set _stats_logger on the hand-built backend,
mirroring __init__, so the kill path's finally has the attribute it expects.

* test: assert torchao override step on normal Linux, not overrides.txt

#6400 moved the torchao dependency override from a fixed pin in overrides.txt to
a torch-matched spec installed via --force-reinstall (_select_torchao_spec), and
turned overrides.txt into a comment-only pointer. It updated the Windows variant
(test_windows_only_includes_overrides) to check for --reinstall, but left
test_normal_linux_includes_overrides asserting overrides.txt is installed, which
no longer happens. Check for the override step (--reinstall) instead, matching
the Windows test.

* test(ui): tolerate ERR_ABORTED on /login re-login in shutdown step

The Shutdown step re-logs in after a CLI password rotation that revoked the prior
token. The SPA auth guard can client-side-redirect mid-navigation against the
stale token, aborting page.goto("/login") with net::ERR_ABORTED. It is a race
(passes on main most of the time). Resolve on domcontentloaded and tolerate the
abort, relying on the password-field wait that follows to confirm we reached
/login, matching the wait_until used by the other navigations in this file.

studio/backend/tests/test_llama_cpp_wait_for_vram_settle.py

@@ -281,6 +281,7 @@ def test_kill_process_records_timestamp_on_actual_kill():
     backend = LlamaCppBackend.__new__(LlamaCppBackend)
     backend._process = None
     backend._healthy = False
+    backend._stats_logger = None  # _kill_process stops it in finally
     backend._stdout_thread = None
     backend._llama_log_fh = None
     backend._last_kill_monotonic = 0.0

tests/python/test_no_torch_filtering.py

@@ -471,11 +471,15 @@ class TestInstallPythonStackSubprocessMock:
     # -- Normal Linux path (NO_TORCH=False, IS_MACOS=False, IS_WINDOWS=False) --
 
     def test_normal_linux_includes_overrides(self):
-        """Normal Linux: overrides.txt IS called."""
+        """Normal Linux: the torchao override step IS called.
+
+        The override step installs a torch-matched torchao spec via
+        --force-reinstall (uv: --reinstall), not overrides.txt directly.
+        """
         cmds = self._capture_install(no_torch = False, is_macos = False, is_windows = False)
-        assert self._cmds_contain_file(
-            cmds, "overrides.txt"
-        ), "overrides.txt should be called on normal Linux"
+        assert any(
+            "--reinstall" in cmd for cmd in cmds
+        ), "torchao override step (--reinstall) should be called on normal Linux"
 
     def test_normal_linux_includes_triton(self):
         """Normal Linux: triton-kernels.txt IS called."""

tests/studio/playwright_chat_ui.py

@@ -1430,7 +1430,16 @@ with sync_playwright() as p:
     # Re-login through the UI with NEW2 so the browser has a valid
     # access token for the /api/shutdown call (the previous one
     # was invalidated by the CLI rotation above).
-    page.goto(f"{BASE}/login")
+    # The CLI rotation left a stale token, so the SPA auth guard can
+    # client-side-redirect mid-navigation and abort this goto with
+    # net::ERR_ABORTED. Resolve on domcontentloaded and tolerate the
+    # abort; the password-field wait below confirms we reached /login.
+    try:
+        page.goto(f"{BASE}/login", wait_until = "domcontentloaded", timeout = 60_000)
+    except Exception as exc:
+        if "ERR_ABORTED" not in str(exc):
+            raise
+        info(f"goto /login aborted ({exc!r}); password-field wait will confirm /login")
     pw_field = page.locator("#password")
     pw_field.wait_for(state = "visible", timeout = 60_000)
     pw_field.fill(NEW2)