Remove old test metrics function

Add processing stats

# Conflicts:
#	bin/kaldi/extract_xvectors.py

+ Download eval dataset
+ Eval script on other voxceleb tasks (-O-clean, -H-clean...)
+ add CONTRIBUTORS.md
+ suppress s3prl warning on other ssl model (but keep warning for wavlm)
+ test custom python loaded model
+ add scoring.sh file to easy score voxceleb tasks

Fix minor bug in sidekit test files creation

Current log:
CCE not working
Rawnet not working

TODO refactor forward to be more simple
TODO non mixed_precision doesn't work!
```
======================================================================
ERROR: test_fastresnet34_smn (__main__.TestFunctions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_xtractor.py", line 121, in test_fastresnet34_smn
    assert xtractor(batch)[1].shape[1] == 256
IndexError: tuple index out of range

======================================================================
ERROR: test_rawnet2_aam (__main__.TestFunctions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_xtractor.py", line 150, in test_rawnet2_aam
    assert isinstance(faker_trainer(xtractor), Xtractor)
  File "tests/test_xtractor.py", line 73, in faker_trainer
    return train_epoch(
  File "/lium/raid01_b/pchampi/lab/sidekit/sidekit/nnet/xvector.py", line 1063, in train_epoch
    loss, no_margin_output = output_tuple
ValueError: too many values to unpack (expected 2)

======================================================================
ERROR: test_rawnet2_cce (__main__.TestFunctions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_xtractor.py", line 157, in test_rawnet2_cce
    assert isinstance(faker_trainer(xtractor), Xtractor)
  File "tests/test_xtractor.py", line 73, in faker_trainer
    return train_epoch(
  File "/lium/raid01_b/pchampi/lab/sidekit/sidekit/nnet/xvector.py", line 1088, in train_epoch
    accuracy += (torch.argmax(no_margin_output.data, 1) == target).sum().cpu()
UnboundLocalError: local variable 'no_margin_output' referenced before assignment

======================================================================
ERROR: test_xvector_cce (__main__.TestFunctions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_xtractor.py", line 142, in test_xvector_cce
    assert xtractor(batch)[1].shape[1] == 256
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/lium/raid01_b/pchampi/lab/sidekit/sidekit/nnet/xvector.py", line 440, in forward
    return self.after_speaker_embedding(x), x
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/modules/container.py", line 141, in forward
    input = module(input)
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/modules/batchnorm.py", line 168, in forward
    return F.batch_norm(
  File "/lium/home/pchampi/lab/sidekit/venv/lib/python3.8/site-packages/torch/nn/functional.py", line 2282, in batch_norm
    return torch.batch_norm(
RuntimeError: running_mean should contain 256 elements not 512

----------------------------------------------------------------------
Ran 13 tests in 140.918s

FAILED (errors=4)
```

Fix bug in csv creation from kaldi