Better characterization of suspend current

[Werner Almesberger]

Andy Green wrote:
> When I see people with bench PSUs [...]

Due to the various differences between a real battery and the average
bench supply (*), we should perhaps define those tests with a battery
and low-resistance contacts.

So for short tests, the test condition would be an initial battery
voltage of N volts, and the final voltage would have to be within a
narrow range of that.

(*) Note: some of the supplies you saw in the lab in Taipei are actually
    extremely fast and should simulate a battery quite accurately. They
    also cost about as much as a good oscilloscope ...

Bench supply tests would still be useful for tracking down leaks and
such.

- Werner