Fixed slating issues caused by NaN samples

Fixes the NOAA APT slanting issues. For some reason the first sample passed to the sink on the Pi will always be NaN. Since arithmetic operations with NaN will also yield NaN this sample 'poisons' the floating average obtained via exponential smoothing, which is used by the code looking for the markers. Comparisons with NaN will always be false, so the code detecting the markers becomes disabled.

This MR introduces code to skip any NaN samples.

See https://community.libre.space/t/noaa-apt-slanting-issues/2264 for the full story.

Please note that this MR does not include the bug fix in !155 (closed), it has been used for testing but should be merged separately.

Also thanks to oldbug and chibill for helping me with the flowgraph poking.