Should remove a suspected overflow in fixed point calculation.

- slower updates (1/3rd and 2 seconds)
- Match precision of info area and live values

... and restart the long-term measurements.

- Total charge
- Voltage min/max
- Max current (in both directions)
- Max power (in both directions)

- LCD pin order
- use 12 MHz crystal
- more stuff I forgot

+ use most recent libopencm3 and fxplib
+ fix the errors resulting from that