AP_NavEKF2: init rngOnGnd to 5cm to avoid div-by-zero

[rmackay9]

This PR is the equivalent of PR #17577 but for EKF2. This issue was reported as part of 4.1 beta testing.

This resolves a crash in SITL by ensuring that the rngOnGnd is never zero and thus avoids a divide-by-zero when fusing optical flow measurements here in AP_NavEKF2_OptFlowFusion.cpp.

The problem is in the lines shown below. "range" can be zero because "rngOnGnd" can be zero which leads to the div-by-zero. We fix this by simply initialising rngOnGnd to 5cm and from then on it can never be set back to zero.

        // calculate range from ground plain to centre of sensor fov assuming flat earth
        ftype range = constrain_ftype((heightAboveGndEst/prevTnb.c.z),rngOnGnd,1000.0f);

        <snip>

        // divide velocity by range  to get predicted angular LOS rates relative to X and Y axes
        losPred[0] =  relVelSensor.y/range;
        losPred[1] = -relVelSensor.x/range;

Testing has been only in SITL but I have checked before and after that this resolves the divide-by-zero. I think from inspection it is also clear that this is a safe change.

[peterbarker]

It's consistent if nothing else :-)

[rmackay9]

Cool, txs and merged!