The issue was the SysTick timer, which was configured by the bootloader to
blink the backlight LEDs, but was not de-configured before booting to the
application.

Without the SysTick timer de-configured, the FreeRTOS tick timer, which relies
on the SysTick timer, fired before FreeRTOS could be properly configured. Since
the FreeRTOS SysTick handler contains some OS-specific functions and data, the
fact that these were not initialized probably led to a segfault.

This issue was fixed in boot.c, by de-configuring the SysTick timer.

Some open-field testing needs to be performed on GPS reception, as the GPS
readouts are ~1 arcsec. off from reference Android app (`My GPS Coordinates').
However, this does translate in 30m accuracy, so it's not a big deal.

Now pending processor unlock to test coordinate format display.

Tested, but not yet working. SW_BL makes loop pass, but no USB connection yet.

Broke one button, attempting to change files accordingly.

Again switching to Fede's branch

Note: the application still doesn't show the coordinates, just UTC
time received from sattellites.

The GPS icon blinks while searching for signal and is on when fixed

Note: still standalone, 'bear'-metal application, not yet hooked
to FreeRTOS

Implemented and tested some utility functions

The subject is the test program for the GPS.

This was actually wrong, the Antenova M10478 GPS module doesn't need a double
ON_OFF pin pulse, it just needs an ~150ms delay from the time its EXT_RST pin
is asserted to the time it can dump out data.

What was happening before is that we didn't have a delay in the gps_init
function before sending the on_off pulse, and this meant that the pulse was
missed by the module, which never turned on.

The added delay fixes this issue and we now need only a single pulse.