Add safety guard against error-prone sensor data

[mwood77]

Cheap pressure sensors are prone to throwing trash data (they get shorted-out easily). We need to programmatically guard against this by setting duty cycle to 0% if this goes above a certain threshold.

Possible solution considerations:

• Verify the mV reading against group head temperature?
• Verify against total uptime & group head temperature
• check span between last (n)number of readings and current. If span is greater than (n), cut power?