Program reported to hang during flash writes

[jynik]

I've seen reports of kalibrate-bladeRF hanging at/prior to the flash write.

I noted that the stop() method appears to be left unimplemented. Assuming stop() gets called prior to the flash write, I suspect this should improve the situation.

Given that the flash accesses must switch the FX3 over to the SPI i/f and disable the 32-bit GPIF, it is absolutely necessary to cleanly shut down the sample stream prior to making flash calls.

It is also important that no attempts to make the FX3 UART (i.e. device control calls) be made while writing flash -- however, libbladeRF's internal device mutex should ensure that.