Fix accept mechanism in tests

[NicolasHug]

This PR addresses #3697 (comment)

I removed the --accept option which doesn't seem to play out really well with pytest, and it's redundant with the already-supported env variable anyway.

Test case below where I delete the pkl file for alexnet, get an expected error, then use the suggested fix in the error message and re-run the test, which passes (sorry it doesn't render super well):

(pt) ➜  vision git:(mungedid) ✗ mv test/expect/ModelTester.test_alexnet_expect.pkl ~
(pt) ➜  vision git:(mungedid) ✗ pytest test/test_models.py -x -k alex --tb=short
============================================================================== test session starts ===============================================================================
platform darwin -- Python 3.8.8, pytest-6.2.2, py-1.10.0, pluggy-0.13.1
rootdir: /Users/nicolashug/dev/vision
plugins: cov-2.11.1
collected 67 items / 66 deselected / 1 selected

test/test_models.py F                                                                                                                                                      [100%]

==================================================================================== FAILURES ====================================================================================
____________________________________________________________________ test_classification_model[dev0-alexnet] _____________________________________________________________________
test/test_models.py:438: in test_classification_model
    ModelTester()._test_classification_model(model_name, input_shape, dev)
test/test_models.py:84: in _test_classification_model
    self.assertExpected(out.cpu(), name, prec=0.1)
test/common_utils.py:126: in assertExpected
    expected_file = self._get_expected_file(name)
test/common_utils.py:109: in _get_expected_file
    raise RuntimeError(
E   RuntimeError: No expect file exists for ModelTester.test_alexnet_expect.pkl in /Users/nicolashug/dev/vision/test/expect/ModelTester.test_alexnet_expect.pkl; to accept the current output, re-run the failing test after setting the EXPECTTEST_ACCEPT env variable. For example: EXPECTTEST_ACCEPT=1 pytest test/test_models.py -k alexnet
============================================================================ short test summary info =============================================================================
FAILED test/test_models.py::test_classification_model[dev0-alexnet] - RuntimeError: No expect file exists for ModelTester.test_alexnet_expect.pkl in /Users/nicolashug/dev/visi...
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! stopping after 1 failures !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
======================================================================== 1 failed, 66 deselected in 1.06s ========================================================================




(pt) ➜  vision git:(mungedid) ✗ EXPECTTEST_ACCEPT=1 pytest test/test_models.py -k alexnet
============================================================================== test session starts ===============================================================================
platform darwin -- Python 3.8.8, pytest-6.2.2, py-1.10.0, pluggy-0.13.1
rootdir: /Users/nicolashug/dev/vision
plugins: cov-2.11.1
collected 67 items / 66 deselected / 1 selected

test/test_models.py .                                                                                                                                                      [100%]

================================================================================ warnings summary ================================================================================
test/test_models.py::test_classification_model[dev0-alexnet]
  /Users/nicolashug/dev/vision/test/common_utils.py:270: RuntimeWarning: The check_jit_scriptable test for AlexNet was skipped. This test checks if the module's results in TorchScript match eager and that it can be exported. To run these tests make sure you set the environment variable PYTORCH_TEST_WITH_SLOW=1 and that the test is not manually skipped.
    warnings.warn(msg, RuntimeWarning)

-- Docs: https://docs.pytest.org/en/stable/warnings.html
================================================================== 1 passed, 66 deselected, 1 warning in 0.93s ===================================================================

[datumbox]

LGTM. I'm always using the env var, so if the pytest does not play well with arguments I don't have issues removing it.

[NicolasHug]

if the pytest does not play well with arguments I don't have issues removing it.

To be perfeclty honest there probably is a way, but since the env is working fine I didn't bother to try