bytes vs str in mig/shared/sandbox.py on experimental with python3

[jonasbardino]

We see log crashes from oneclick like

UnicodeDecodeError: 'utf-8' codec can't decode byte 0x8f in position 0: invalid start byte
Traceback (most recent call last):
  File "/home/mig/mig/shared/cgiscriptstub.py", line 185, in run_cgi_script_possibly_with_cert
    (out_obj, (ret_code, ret_msg)) = main(client_id, user_arguments_dict)
  File "/home/mig/mig/shared/functionality/oneclick.py", line 72, in main
    (status, result) = get_resource(client_id, configuration, logger)
  File "/home/mig/mig/shared/sandbox.py", line 371, in get_resource
    sandboxkey = hexlify(open('/dev/urandom').read(32))
  File "/usr/lib64/python3.9/codecs.py", line 322, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)

apparently because the plain open() defaults to text-mode and therefore tries to parse the raw bytes as utf8 on python3

[jonasbardino]

The follow-up uncovered a bunch of related issues in other modules preventing it from working with the initial fix.
We generally should not rely on the old fragile open(/dev/urandom).read(N) calls but stick with the modern os.urandom(N) instead. Handled in pending PR163.

[jonasbardino]

Solved with the merge of PR163.