I found 3 problems with the GS modeling in flightgear (all my fault originally

I found 3 problems with the GS modeling in flightgear (all my fault originally
I believe.) :-)

- The height of the navaid was not being properly converted to meters
  before being used in our internal calculations.  This caused the GS
  to be placed too high.

- I was using the wrong trig function to calculate the current approach
  angle of the aircraft.  The distance to the GS source is the euclidean
  point to point distance and represents the hypotenuse (not the ground
  distance) so I need to use asin() rather than atan() to calculate the
  angle.

- I was calculating distance directly to the GS source, rather than
  taking into consideration that the GS transmitter projects a plane,
  so I need to take the distance to the line where that plane intersectso
  the ground.  Previously, the way I modeled my distance calculation, the
  GS transmitter effectively formed a 3 degree cone from the source.  The GS
  transmitter is usually placed a 100 meters or so off the runway edge so
  the cone model could never bring you in to the touch down point precisely.

With these changes, the GS will bring you in precisely to the touchdown
point as defined in the default.ils.gz file (it wouldn't before.)  The only
issue that remains is that it will bring you in to the elevation defined
in the ILS database, which doesn't necessarily match the DEM/SRTM terrain
at that point.  Still on average, this will be a big improvement until we
can do a better job of getting the runway end elevations nailed correctly.