use charset_normalizer

[brc-dd]

Currently, all the text-like files are loaded as utf-8. But this causes issues, for example, with csv files created on Windows in Japanese. Those might be encoded as CP932. https://github.com/jawah/charset_normalizer can help with this issue. Currently it throws this:

markitdown._markitdown.FileConversionException: Could not convert 'test.csv' to Markdown. File type was recognized as ['.csv']. While converting the file, the following error was encountered:

Traceback (most recent call last):
  File "/Users/brc-dd/foo/.venv/lib/python3.12/site-packages/markitdown/_markitdown.py", line 1041, in _convert
    res = converter.convert(local_path, **_kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/brc-dd/foo/.venv/lib/python3.12/site-packages/markitdown/_markitdown.py", line 166, in convert
    text_content = fh.read()
                   ^^^^^^^^^
  File "<frozen codecs>", line 322, in decode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x91 in position 1: invalid start byte

chardet doesn't work with such files. It detects them as Shift JIS while Windows uses an extended version of Shift JIS (CP932).