`Write` keyword fails, if terminal contains non-utf8 characters

[uruun]

The Read Until With Encoding Errors On Strict test is failing with UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfa in position 88: invalid start byte (the "position" can differ). Turns out that the corrupted output of the file is leaking to the Write keyword call. It makes Write fail depending on timing, if the output following Write is corrupted. Issue introduced in #431

def _read_until(self, matcher, expected, timeout=None):
     timeout = TimeEntry(timeout) if timeout else self.config.get('timeout')
     max_time = time.time() + timeout.value
     while time.time() < max_time:
         undecoded = self._single_complete_read_to_buffer(max_time)  # <<<<< Reading into corrupted text
         if undecoded:
             self._receive_buffer += undecoded.decode(  # <<<<< Exception
                 self.config.encoding, "ignore"
             )
         match = matcher(self._receive_buffer)
         if match:
             if hasattr(match, "end"):
                 end = match.end()
             else:
                 end = self._receive_buffer.index(expected) + len(expected)
             output = self._receive_buffer[0:end]
             self._receive_buffer = self._receive_buffer[end:]
             return output
     output = self._receive_buffer
     self._receive_buffer = ""
     raise SSHClientException(f"No match found for '{expected}' in {timeout}\nOutput:\n{output}.")

If the read after Write reads beyond the written text into the corrupted part, the

undecoded = self._single_complete_read_to_buffer(max_time)

will read until timeout and the undecoded.decode will fail. The fix might be to try and decode as much of the beginning of the undecoded part as possible and then match.