ubxtool and python 3

[Greg Troxel]

Some good news - I disabled SBAS (M8T, I am pretty sure), and saw on
xgps "3D" (no DIFF), and then when I enabled ntrip saw DIFF, and
convinced myself the positioning was much tighter.  Between reading the
code, tcpdump, web pages, and your hints, I am now convinced I am
actually getting differential corrections over NTRIP.

I tried to run ubxtool via gpsd with ntrip, and got an error about
mulitple devices.  I would sort of expect it can figure out which is a
ublox and which isn't, and use the sole one, but that's another story.
I then read the man page to select one, and:

  $ ubxtool  -p  CFG-GNSS localhost:2947:/dev/dtyU0
  ubxtool: poll CFG-GNSS
  Traceback (most recent call last):
    File "/usr/pkg/bin/ubxtool", line 7066, in <module>
      command["command"](gps, command["opt"])
    File "/usr/pkg/bin/ubxtool", line 6239, in send_poll
      gps_model.gps_send(m_data[0], m_data[1], m_data[2:])
    File "/usr/pkg/bin/ubxtool", line 5699, in gps_send
      self.gps_send_raw(m_all)
    File "/usr/pkg/bin/ubxtool", line 5704, in gps_send_raw
      io_handle.ser.write(m_all)
    File "/usr/pkg/bin/ubxtool", line 6811, in write_gpsd
      data_out = b'?DEVICE={"path":"' + opts['target']['device'] + b'",'
  TypeError: can't concat str to bytes

think I have run into a python 3 bug (3.7.5 here).