Separator failing with "Trying to set the stream twice. This isn't supported. "

- [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
Anyone have a working Windows 10 version of spleeter - seems dependencies have changed and I cant find a working combination.
Tried python 3.6-3.10
Tried 2stems, 4stems, 5stems
Tensorflow - various versions up to 2.10 - even to the latest.
Windows 10. Using Nvida GPU but disabled or not using in the code.
Tried the SpleeterGUI .zip as well .same error "Trying to set the stream twice. This isn't supported."
Tried minimal requirements.txt, tried supplied requirements.txt
Tried other similar packages using their requirements.txt. . like TubeSplitter. Same errors.
Anyone provide a working requirements.txt file?
(from what I can tell this repo isnt really supported anymore but hoping others can chime in)

```
WARNING:tensorflow:From C:\Users\daryl\.virtualenvs\Spleeter\lib\site-packages\spleeter\separator.py:146: calling DatasetV2.from_generator (from tensorflow.python.data.ops.dataset_ops) with output_types is deprecated and will be removed in a future version.
Instructions for updating:
Use output_signature instead
WARNING:tensorflow:From C:\Users\daryl\.virtualenvs\Spleeter\lib\site-packages\spleeter\separator.py:146: calling DatasetV2.from_generator (from tensorflow.python.data.ops.dataset_ops) with output_shapes is deprecated and will be removed in a future version.
Instructions for updating:
Use output_signature instead
WARNING:tensorflow:From C:\Users\daryl\.virtualenvs\Spleeter\lib\site-packages\keras\layers\normalization\batch_normalization.py:514: _colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.
Instructions for updating:
Colocations handled automatically by placer.
2024-05-11 16:38:30.243361: F tensorflow/core/common_runtime/gpu/gpu_cudamallocasync_allocator.cc:390] Trying to set the stream twice. This isn't supported.
```

Code clip from AnNOTEator github.

```
waveform, _ = audio_adapter.load(
path,
offset=drum_start if drum_start is not None else None,
duration=drum_end-drum_start if drum_end is not None else None,
sample_rate=sample_rate
)

prediction = separator.separate(waveform)
```

Start with the TensorFlow failure reported from spleeter/separator.py:146 and compare the Windows 10 environment details with the repository's requirements.txt. Reproduce the AnNOTEator call using separator.separate, then determine whether a supported dependency combination avoids the GPU allocator error; done means a documented working setup or a confirmed project-level fix.

Written by the indexing model from the issue text.