consider revising vbat supply

[mck1117]

2 problems observed:

1. the zener diode based shunt regulator isn't particularly accurate, since the current passed is relatively low. The chosen zener is specified at 5mA - while the stm32's vbat pin draws 3 orders of magnitude less current. This is probably actually okay, since the STM32's vbat pin operates at 1.65-3.3 volts.
2. Since it is a simple zener-based shunt, the power consumption is very high for a standby circuit. Typical industry standby current for an ECU is on the order of 50-100uA. The current standby circuit draws several times more than this: 280uA estimated, 296uA measured on h72b at 12.5 volts supply.

Both the accuracy and power consumption can be solved by using a standard linear regulator instead, essentially all of which have a quiescent current consumption far less than 100uA. The STM32 only draws 5-10uA for the RTC, a rounding error compared to the shunt.

Switching to a proper regulator will also allow for eliminating the relatively large and expensive 100uF tantalum capacitor.