@Fungus
You are a hopeless case, mate. I do wonder why did you even bother posting here when you know that your problem is that stupid FTDI driver (despite having no evidence ...), refuse to actually cooperate with people who are trying to help you out to debug your issue, when your lack of understanding is pointed out to you, you get defensive instead of trying to understand what is being said and only think we have you for an idiot ...
I will acknowledge Fungus' point that the driver has to be at fault to some degree, and doing something bad when it could do something better, such as not returning control the OS or similar. This belief is also reinforced by the fact that simply updating the driver (assuming no other changes were made) seems to have made the problem less likely to happen.
However, I believe there's a bunch of things that he could have tested, such as what I suggested above about disconnecting one serial wire at a time and see if the problem still happens. Examples:
Does the problem still happen if disconnecting only the TX line from the Arduino to the serial adapter, but keeping the RX line, and optionally ground wire, connected ? If so, then by a good first approximation the problem is with something transferred over that line, but there would be further steps to prove this.
Does the problem still happen when disconnecting both serial lines between the wireless module and the Arduino? If yes, then, this has to happen through some out-of-band communcation, such as radiated or conducted interference, or a short-circuit/power overload.
But there's still the off-chance that the FTDI driver is not the sole factor to blame, but that the problem comes from having the Arduino's serial driver loaded at the same time as the FTDI driver. Does the problem still happen if you reset with the FTDI module plugged in to USB, but fully disconnected from the Arduino? If so the likely culprit is a driver combination, but there are further steps to prove this beyond a reasonable doubt if that's where things are pointing.
All of the above steps would of course have to be done with the old driver (which can still be rolled back to the old version in Windows, btw.)
I'm seeing an inability, or more likely unwillingness, on Fungus' part to perform such steps and draw conclusions. I can understand not wanting to sink time into this, but in the same time, this is a missed opportunity for knowledge, which frustrates me.