Describe the bug
As described in the README, the library provides an interface to retrieve normalized text from an input stream.

In the case of some UTF7 strings with byte-order-marks, despite the library often correctly identifying the encoding as UTF7, the normalized/decoded text results are unreliable.

To Reproduce
Inline Python bytestring examples below:

from charset_normalizer import from_bytes
str(from_bytes('\ufeff.testing'.encode('utf-7')).best())  # works as expected
str(from_bytes('\ufeff-testing'.encode('utf-7')).best())  # returns extraneous leading minus-symbol
str(from_bytes('\ufeff+testing'.encode('utf-7')).best())  # returns extraneous b64-ish prefix

Expected behavior
The above would ideally return:

'.testing'
'-testing'
'+testing'

Actual behaviour
The results for v3.4.6 are:

'.testing'
'--testing'
'AKw-testing'

Logs
If applicable, add console outputs to help explain your problem.

Desktop (please complete the following information):

• OS: Debian GNU/Linux (testing / forky)
• Python version: 3.13.12
• Package version: 3.4.6

Additional context
Encountered after exploring the issue described in #716.