[Bug] NotImplementedError when running Docker `deezer/spleeter:conda`

- [x] I didn't find a similar issue already open.
- [x] I read the documentation (README AND Wiki)
- [x] I have installed FFMpeg
- [x] My problem is related to Spleeter only, not a derivative product (such as Webapplication, or GUI provided by others)

## Description

When running docker image `deezer/spleeter:conda` I encounter a `NotImplementedError` when trying to seperate an audio file.

## Step to reproduce

1. Executed the following command: `docker run -v $(pwd)/output:/output -v $(pwd)/input:/input deezer/spleeter:conda separate -o /output /input/audio_example.mp3`

## Output

```bash
Traceback (most recent call last):
File "/opt/conda/bin/spleeter", line 11, in
sys.exit(entrypoint())
File "/opt/conda/lib/python3.7/site-packages/spleeter/__main__.py", line 250, in entrypoint
spleeter()
File "/opt/conda/lib/python3.7/site-packages/typer/main.py", line 214, in __call__
return get_command(self)(*args, **kwargs)
File "/opt/conda/lib/python3.7/site-packages/click/core.py", line 829, in __call__
return self.main(*args, **kwargs)
File "/opt/conda/lib/python3.7/site-packages/click/core.py", line 782, in main
rv = self.invoke(ctx)
File "/opt/conda/lib/python3.7/site-packages/click/core.py", line 1259, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/opt/conda/lib/python3.7/site-packages/click/core.py", line 1066, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/opt/conda/lib/python3.7/site-packages/click/core.py", line 610, in invoke
return callback(*args, **kwargs)
File "/opt/conda/lib/python3.7/site-packages/typer/main.py", line 497, in wrapper
return callback(**use_params) # type: ignore
File "/opt/conda/lib/python3.7/site-packages/spleeter/__main__.py", line 131, in separate
synchronous=False,
File "/opt/conda/lib/python3.7/site-packages/spleeter/separator.py", line 382, in separate_to_file
sources = self.separate(waveform, audio_descriptor)
File "/opt/conda/lib/python3.7/site-packages/spleeter/separator.py", line 325, in separate
return self._separate_librosa(waveform, audio_descriptor)
File "/opt/conda/lib/python3.7/site-packages/spleeter/separator.py", line 263, in _separate_librosa
outputs = self._get_builder().outputs
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 318, in outputs
self._build_outputs()
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 501, in _build_outputs
self._outputs = self.masked_stfts
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 342, in masked_stfts
self._build_masked_stfts()
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 465, in _build_masked_stfts
for instrument, mask in self.masks.items():
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 336, in masks
self._build_masks()
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 449, in _build_masks
instrument_mask = self._extend_mask(instrument_mask)
File "/opt/conda/lib/python3.7/site-packages/spleeter/model/__init__.py", line 420, in _extend_mask
extension_row = tf.zeros((mask_shape[0], mask_shape[1], 1, mask_shape[-1]))
File "/opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py", line 201, in wrapper
return target(*args, **kwargs)
File "/opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/array_ops.py", line 2747, in wrapped
tensor = fun(*args, **kwargs)
File "/opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/array_ops.py", line 2794, in zeros
output = _constant_if_small(zero, shape, dtype, name)
File "/opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/array_ops.py", line 2732, in _constant_if_small
if np.prod(shape) < 1000:
File "<__array_function__ internals>", line 6, in prod
File "/opt/conda/lib/python3.7/site-packages/numpy/core/fromnumeric.py", line 3031, in prod
keepdims=keepdims, initial=initial, where=where)
File "/opt/conda/lib/python3.7/site-packages/numpy/core/fromnumeric.py", line 87, in _wrapreduction
return ufunc.reduce(obj, axis, dtype, out, **passkwargs)
File "/opt/conda/lib/python3.7/site-packages/tensorflow/python/framework/ops.py", line 848, in __array__
" a NumPy call, which is not supported".format(self.name))
NotImplementedError: Cannot convert a symbolic Tensor (strided_slice_4:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported
```

## Environment

| | |
| ----------------- | ------------------------------- |
| OS | Linux |
| Installation type | Docker |
| RAM available | 4GB |
| Hardware spec | N/A |

Reproduce the failure with the documented docker run command and inspect spleeter/model/__init__.py, especially _extend_mask and _build_masks, using the traceback as the entry point. Confirm the Docker image can separate the example MP3 without the symbolic-Tensor NumPy error, then verify that output stems are written to /output.

Written by the indexing model from the issue text.