fix: wrongly fallback to low memory model when model corrupted

[JackyHe398]

fix of issue #19. The model now should normally through an error when the model corrupted and fallback to low memory model when memory not enough.

test.py:

import logging
logging.basicConfig(level=logging.DEBUG, format='%(levelname)s:%(name)s:%(message)s')

from src import fast_langdetect
fast_langdetect.infer._default_detector = fast_langdetect.infer.LangDetector(fast_langdetect.infer.LangDetectConfig(cache_dir="/tmp/fasttext-langdetect"))

import os, sys, resource
limit = <limit in MiB>* 1024 * 1024
resource.setrlimit(resource.RLIMIT_AS, (limit, limit))

print("Testing different behavior between MemoryError and model corruption:")
result = fast_langdetect.detect('Hello world', low_memory=False)
print('Result:', result)

result when model corrupted:

(GPTSoVITS) hekh@ljz-MS-7D76:/tmp/fasttext-langdetect$ /home/hekh/miniconda3/envs/GPTSoVITS/bin/python /home/hekh/code/fast-langdetect/test.py
Testing different behavior between MemoryError and model corruption:
Traceback (most recent call last):
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 192, in _load_unix
    return fasttext.load_model(str(model_path))
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/hekh/miniconda3/envs/GPTSoVITS/lib/python3.11/site-packages/fasttext/FastText.py", line 82, in load_model
    return _FastText(model_path=path)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/hekh/miniconda3/envs/GPTSoVITS/lib/python3.11/site-packages/fasttext/FastText.py", line 29, in __init__
    self.f.loadModel(model_path)
ValueError: /tmp/fasttext-langdetect/lid.176.bin has wrong file format!

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/hekh/code/fast-langdetect/test.py", line 12, in <module>
    result = fast_langdetect.detect('Hello world', low_memory=False)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 444, in detect
    return _default_detector.detect(text, low_memory=low_memory)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 343, in detect
    model = self._get_model(low_memory)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 317, in _get_model
    model = self._model_loader.load_with_download(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 130, in load_with_download
    return self.load_local(model_path)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 124, in load_local
    return self._load_unix(model_path)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ljz/code/fast-langdetect/src/fast_langdetect/infer.py", line 197, in _load_unix
    raise DetectError(f"fast-langdetect: Failed to load model: {e}")
src.fast_langdetect.infer.DetectError: fast-langdetect: Failed to load model: /tmp/fasttext-langdetect/lid.176.bin has wrong file format!

result when memory not enough but model normal:

(GPTSoVITS) hekh@ljz-MS-7D76:/tmp/fasttext-langdetect$ /home/hekh/miniconda3/envs/GPTSoVITS/bin/python /home/hekh/code/fast-langdetect/test.py
Testing different behavior between MemoryError and model corruption:
INFO:src.fast_langdetect.infer:fast-langdetect: Falling back to low-memory model...
Result: {'lang': 'en', 'score': 0.1682594120502472}

[sudoskys]

Thank you for your contribution! It means a lot.