Run non-TF fork tests before TF fork tests so the TF wedge can't mask them

The pt/onnx fork tests landed in the same serial -n 0 fork lane as the
TF fork tests. Because a fork-after-TensorFlow wedge kills the whole -n 0
process (that is why the lane is -n 0), and perch_v2 sorts before v3_0,
the intermittent TF wedge killed the lane before the pt/onnx fork tests
ran -- so the very tests meant to prove fork works without TensorFlow
never executed (observed in run 30630301430: fork[pt]/fork[onnx] never
ran; forkserver/spawn passed in the general lane).

Give the deadlock-free non-TF fork tests a "fork_nontf" marker and run
them first in their own process, ahead of the TF fork tests. A wedge in
the TF lane can no longer stop them from recording a result. They keep
the "fork" marker too, so the parallel general lane still excludes them
(forking after TF is loaded there would deadlock).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: Josef-Haupt

pyproject.toml

@@ -208,6 +208,7 @@ markers = [
   "repro: tests requiring exact package versions to reproduce results",
   "no_tf: tests for the TensorFlow-free surface, run when TF is absent (e.g. Python 3.14)",
   "fork: tests that force the fork start method; run in a dedicated serial (-n 1) lane in a fresh process so a fork-after-TensorFlow deadlock cannot wedge the parallel general-phase workers (the library itself defaults to spawn, see birdnet.core.start_method)",
+  "fork_nontf: fork tests for non-TensorFlow backends (pt/onnx), which are deadlock-free; run in their own serial invocation BEFORE the TF fork tests so the TF tests' intermittent fork-after-TensorFlow wedge cannot stop them from proving fork works without TensorFlow (also carry `fork` so the general lane still excludes them)",
 ]
 
 [tool.ruff]
@@ -285,7 +286,12 @@ commands =
   # with the worker, and the orphaned child keeps the worker's execnet socket open
   # so the controller sits blind until the watchdog's idle deadline. In-process,
   # the timeout report lands in the log and the run ends at the test timeout.
-  pytest -m "not repro and not load_model and fork and not litert and not gpu" -n 0
+  # The non-TF (pt/onnx) fork tests run first in their own invocation: they are
+  # deadlock-free, and a fresh process means the TF fork tests' intermittent wedge
+  # (which kills the rest of its own -n 0 run) cannot stop them from recording a
+  # result. See the `fork_nontf` marker.
+  pytest -m "not repro and not load_model and fork and fork_nontf and not litert and not gpu" -n 0
+  pytest -m "not repro and not load_model and fork and not fork_nontf and not litert and not gpu" -n 0
   pytest -m "not repro and not load_model and fork and litert and not gpu" -n 0
   # build and check package
   # note `python -m build` uses the isolated build folder which results in import error:
@@ -302,8 +308,11 @@ commands =
   pytest -m "not load_model and (not litert and not gpu and not fork)" -n auto
   # run litert tests including repro tests
   pytest -m "not load_model and litert and not fork" -n auto
-  # fork tests isolated + serial (see the `fork` marker and conftest.py)
-  pytest -m "not load_model and fork and not litert and not gpu" -n 0
+  # fork tests isolated + serial (see the `fork` marker and conftest.py).
+  # Non-TF (pt/onnx) fork tests run first in their own process so the TF fork
+  # tests' intermittent wedge cannot stop them recording a result (see fork_nontf).
+  pytest -m "not load_model and fork and fork_nontf and not litert and not gpu" -n 0
+  pytest -m "not load_model and fork and not fork_nontf and not litert and not gpu" -n 0
   pytest -m "not load_model and fork and litert and not gpu" -n 0
   # gpu tests are not supported to be run with repro environment
   pyproject-build -o dist/
@@ -328,8 +337,11 @@ commands =
   pytest -m "not repro and not load_model and (not litert and not gpu and not fork)" -n auto --cov=src/birdnet --cov-append --cov-report=
   pytest -m "not repro and not load_model and litert and not fork" -n auto --cov=src/birdnet --cov-append --cov-report=
   pytest -m "not repro and not load_model and gpu" -n 1 --cov=src/birdnet --cov-append --cov-report=
-  # fork tests isolated + serial (see the `fork` marker and conftest.py)
-  pytest -m "not repro and not load_model and fork and not litert and not gpu" -n 0 --cov=src/birdnet --cov-append --cov-report=
+  # fork tests isolated + serial (see the `fork` marker and conftest.py).
+  # Non-TF (pt/onnx) fork tests run first in their own process so the TF fork
+  # tests' intermittent wedge cannot stop them recording a result (see fork_nontf).
+  pytest -m "not repro and not load_model and fork and fork_nontf and not litert and not gpu" -n 0 --cov=src/birdnet --cov-append --cov-report=
+  pytest -m "not repro and not load_model and fork and not fork_nontf and not litert and not gpu" -n 0 --cov=src/birdnet --cov-append --cov-report=
   pytest -m "not repro and not load_model and fork and litert and not gpu" -n 0 --cov=src/birdnet --cov-append --cov-report=
   coverage report
 """

src/birdnet_tests/acoustic_models/v3_0/model_py/test_predict/test_start_method_v3_0.py

@@ -69,6 +69,7 @@ def _run_two_parallel_sessions(backend: _Backend) -> None:
 # it moves from TensorFlow to PyTorch, so these lanes must stay green; they are the
 # coverage that proves fork works for everything except TensorFlow.
 @pytest.mark.fork
+@pytest.mark.fork_nontf
 @pytest.mark.parametrize("backend", ["pt", "onnx"])
 def test_twice_two_sessions_parallel_processes_fork(backend: _Backend) -> None:
   # macOS is skipped to match the perch/v2.4 fork tests: forking the