Ground Level Estimator becomes sticky as sample count grows

[Elan456]

In the GLE, sample count can grow to massive values. If the rocket is prepped somewhere for an hour and then moved to the launch pad, the ground level won't be able to shift much because the sample count is so large.

I think we should switch to an exponential moving average, so that new samples always have a weight of alpha regardless of long the flight computer has been turned on.

alt = alpha * new_alt_sample + (previous_alt * (1-alpha))

Also launchDeteched() is spelled wrong