Remove torchdata dependency from package and from CI

[NicolasHug]

This PR removes the dependency on torchdata:

• from the torchtext package (setup.py / requirements.txt, etc.)
• from the CI: all unittest and build jobs.

The files that still have references to torchdata after this PR are:

(base) ➜  text git:(remove_torchdata) ✗ git grep -i -l torchdata 
CONTRIBUTING_DATASETS.md
README.rst
docs/source/datasets.rst
examples/torcharrow/README.md
examples/tutorials/sst2_classification_non_distributed.py
examples/tutorials/t5_demo.py
test/torchtext_unittest/datasets/common.py
test/torchtext_unittest/datasets/test_agnews.py
test/torchtext_unittest/datasets/test_amazonreviews.py
test/torchtext_unittest/datasets/test_cc100.py
test/torchtext_unittest/datasets/test_cnndm.py
test/torchtext_unittest/datasets/test_cola.py
test/torchtext_unittest/datasets/test_conll2000chunking.py
test/torchtext_unittest/datasets/test_dbpedia.py
test/torchtext_unittest/datasets/test_enwik9.py
test/torchtext_unittest/datasets/test_imdb.py
test/torchtext_unittest/datasets/test_iwslt2016.py
test/torchtext_unittest/datasets/test_iwslt2017.py
test/torchtext_unittest/datasets/test_mnli.py
test/torchtext_unittest/datasets/test_mrpc.py
test/torchtext_unittest/datasets/test_multi30k.py
test/torchtext_unittest/datasets/test_penntreebank.py
test/torchtext_unittest/datasets/test_qnli.py
test/torchtext_unittest/datasets/test_qqp.py
test/torchtext_unittest/datasets/test_rte.py
test/torchtext_unittest/datasets/test_sogounews.py
test/torchtext_unittest/datasets/test_squads.py
test/torchtext_unittest/datasets/test_sst2.py
test/torchtext_unittest/datasets/test_stsb.py
test/torchtext_unittest/datasets/test_udpos.py
test/torchtext_unittest/datasets/test_wikitexts.py
test/torchtext_unittest/datasets/test_wnli.py
test/torchtext_unittest/datasets/test_yahooanswers.py
test/torchtext_unittest/datasets/test_yelpreviews.py
torchtext/datasets/ag_news.py
torchtext/datasets/amazonreviewfull.py
torchtext/datasets/amazonreviewpolarity.py
torchtext/datasets/cc100.py
torchtext/datasets/cnndm.py
torchtext/datasets/cola.py
torchtext/datasets/conll2000chunking.py
torchtext/datasets/dbpedia.py
torchtext/datasets/enwik9.py
torchtext/datasets/imdb.py
torchtext/datasets/iwslt2016.py
torchtext/datasets/iwslt2017.py
torchtext/datasets/mnli.py
torchtext/datasets/mrpc.py
torchtext/datasets/multi30k.py
torchtext/datasets/penntreebank.py
torchtext/datasets/qnli.py
torchtext/datasets/qqp.py
torchtext/datasets/rte.py
torchtext/datasets/sogounews.py
torchtext/datasets/squad1.py
torchtext/datasets/squad2.py
torchtext/datasets/sst2.py
torchtext/datasets/stsb.py
torchtext/datasets/udpos.py
torchtext/datasets/wikitext103.py
torchtext/datasets/wikitext2.py
torchtext/datasets/wnli.py
torchtext/datasets/yahooanswers.py
torchtext/datasets/yelpreviewfull.py
torchtext/datasets/yelpreviewpolarity.py

The torchtext.datasets namespace still relies on torchdata but that's OK: users who still need it can just install torchdata manually. Similarly I did not update/remove the tests/datasets/* files and instead just ignored all the dataset tests in pytest.ini.

I will be submitting follow-up PRs to also address the user-facing docs, raise proper warnings, etc.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/text/2241

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 19 New Failures, 5 Unrelated Failures

As of commit 9e13774 with merge base 52c0d85 ():

NEW FAILURES - The following jobs have failed:

• Bandit / build (gh)
  Process completed with exit code 1.
• Build Linux Conda / pytorch/text / upload / conda-py3_8-cpu (gh)
  Unable to find any artifacts for the associated workflow
• Build Linux Wheels / pytorch/text / upload / manywheel-py3_8-cpu (gh)
  Unable to find any artifacts for the associated workflow
• Build M1 Wheels / pytorch/text / upload / wheel-py3_8-cpu (gh)
  Unable to find any artifacts for the associated workflow
• Build Windows Conda / pytorch/text / upload / conda-py3_8-cpu (gh)
  Unable to find any artifacts for the associated workflow
• Build Windows Wheels / pytorch/text / upload / wheel-py3_8-cpu (gh)
  Unable to find any artifacts for the associated workflow
• CodeQL / build (gh)
  ModuleNotFoundError: No module named 'torch'
• cron / nightly / validate-binaries / linux-manywheel-3.8-rocm5.7 / linux-manywheel-3.8-rocm5.7 (gh)
  RuntimeError: Command docker exec -t 9cc9aee839ef2160e7a078c8721d6b56eddbe6f84613ec738bb0ca403017b592 /exec failed with exit code 1
• cron / nightly / validate-binaries / linux-manywheel-3.8-rocm6.0 / linux-manywheel-3.8-rocm6.0 (gh)
  RuntimeError: Command docker exec -t b4c59739469ca3cd8d7a1ee09148c751eb6eed08af365234a9366f2767a2fdae /exec failed with exit code 1
• cron / nightly / validate-binaries / macos-conda-3.8-cpu / macos-arm64-conda-3.8-cpu (gh)
  RuntimeError: Command bash /Users/ec2-user/runner/_work/_temp/exec_script failed with exit code 1
• cron / nightly / validate-binaries / macos-conda-3.8-cpu / macos-conda-3.8-cpu (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 1
• Integration Test / tests (3.8) / linux-job (gh)
  RuntimeError: Command docker exec -t 92bd2fbf6c43ea828181aece0404e537bd15a9398ed866bb8626f491aa164c2f /exec failed with exit code 1
• Unit-tests on Linux CPU / tests (3.10) / linux-job (gh)
  RuntimeError: Command docker exec -t 537a58035017a0bb75f9972a544337e74bfd93e344eb5006bbbac84bbf985d77 /exec failed with exit code 1
• Unit-tests on Linux CPU / tests (3.8) / linux-job (gh)
  RuntimeError: Command docker exec -t 04f356f207df4729f1a6c0e19076f86b668e60ac3ca594331a1d2c6d3288ab31 /exec failed with exit code 1
• Unit-tests on Linux CPU / tests (3.9) / linux-job (gh)
  RuntimeError: Command docker exec -t 5cd692a9adfbaed8e7289af58540821d2f156ba487f0889e73407872c3a6da46 /exec failed with exit code 1
• Unit-tests on Linux GPU / tests (3.8, 11.7) / linux-job (gh)
  RuntimeError: Command docker exec -t f748028d3a04c4df1a10a38807c2f8264246395cb2e1e5d2c4588c413a5b1cf3 /exec failed with exit code 1
• Unit-tests on Windows CPU / tests (3.10) / windows-job (gh)
  ImportError: DLL load failed while importing pywintypes: The specified module could not be found.
• Unit-tests on Windows CPU / tests (3.8) / windows-job (gh)
  ImportError: DLL load failed while importing pywintypes: The specified module could not be found.
• Unit-tests on Windows CPU / tests (3.9) / windows-job (gh)
  ImportError: DLL load failed while importing pywintypes: The specified module could not be found.

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• Build M1 Conda / pytorch/text / upload / conda-py3_8-cpu (gh)
• cron / nightly / validate-binaries / windows-conda-3.8-cpu / windows-conda-3.8-cpu (gh)
  Process completed with exit code 1.
• Unit-tests on Macos CPU / tests (3.10) / macos-job (gh)
  thinc/backends/numpy_ops.cpp:5948:3: error: no member named 'use_tracing' in '_PyCFrame'
• Unit-tests on Macos CPU / tests (3.8) / macos-job (gh)
  thinc/backends/numpy_ops.cpp:5948:3: error: no member named 'use_tracing' in '_PyCFrame'
• Unit-tests on Macos CPU / tests (3.9) / macos-job (gh)
  thinc/backends/numpy_ops.cpp:5948:3: error: no member named 'use_tracing' in '_PyCFrame'

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[atalman]

Maybe we could include 1-2 basic torchtext functionalities tests, here to see if they work. This can be done in followup pr