1.) Updating the firmware is only possible when connected via USB.

2.) With such a huge different it is more likely that the sensor is damaged. There is an offset function which allows you to change the temperature in the range of up to +/-6°C.

3.) Flow rate would be very high with this and more than OK - sounds almost unrealistic high.

1) Yes, USB only for updating anything aquacomputer make as far as I know. It would be nice if this could be done over aquabus in future, but I don't know if that will be possible.

2) You can calibrate the sensors with the PC off, that's what I do to get everything as accurate as possible. Shut down, let everything equalise in temperature for a long time (overnight or more), then plug a laptop into the aquaero using an internal->external USB cable, install aquasuite on the laptop and you can change the calibration values to get them all reading the same. No power to the aquaero is needed to do this other than what it gets from the USB connection to the laptop. I agree that being nearly 6c out sounds like a bad sensor, but you could try calibrating it then see if it tracks realistically with the system on when the water warms up - if it does, it's probably OK, or at least close enough.

3) That calibration value sounds fine, and that flow does sound realistic to me. I don't know if Shoggy is familiar with MCP35x but they are certainly very powerful - more than anything other than an Iwaki. I had a single MCP35x running CPU block, 3x full cover GPU blocks, 1 radiator and 10M of 1/2" ID hose (external radbox) and I was getting 1GPM indicated which also sounded right to me.

I'm not using the aquacomputer ones, I'm using one G1/4 plug type in the res and a Phobya inline type one coming out of my rad. I think they are quite accurate - after calibrating them in the way I described, I have poked around testing things with an IR thermometer, and also compared the "power dissipated" calculated by the aquaero (using both water temperatures and the flow measurement) to the power being supplied by the PSU and they seem accurate.

If a second sensor gives you a similar or same problem, I'd suspect the sensor channel on the aquaero and try a different channel.

I tried to calibrate my temp sensors. 6 inline/5 blade types. I put them inside a platic baggie with the inline sensors screwed into each other, and held the bag down inside an ice water bath until temps stopped changinging + 30 min. I got:2.9/2.6/2.8/2.7/2.5/2.5 with inline and 2.0/1.6/1.6/1.2/1.5 with the blade type. The ice cubes kept freezing together so the bath was close to Zero C.
If I had been smart, I would have changed the sensors to a different sensor connector on the AQ5 and found out how sensitive the connectors are in different places on the unit. I didn't.
I was going to check all the sensors at "normal" temps, but the Boss wanted Her ice chest, and my computer mess off her table.
These things are just resistors that change resistance with temperature changes, with a nominal valuse of 10 kohms. They are made very inexpensively and I personally think with a wide variance in values. I do not care if they are +/- 1 degree Centegrade or more, I just want them all close.
I also believe that the physical connection to the AQ5 is critical. Bad connection = bad reading.
Hope this helps a little bit.