Add optional mounting orientation argument to rendering (and perhaps calibration)

[russell-taylor]

The presence of the IMU exposes the fiction of rotating the cameras in the helicopter coordinate system to present that an upside-down camera is right-side-up.

To get appropriate camera-motion correction when the camera is rotating, we'll need to correct both the viewing direction and the differential rotation so that they happen in the same space.

If we also do this with the calibration code, then we won't need to flip it once it has been calibrated.

[russell-taylor]

Why does running ASDP_Render_Module with --lockRotation keep things properly aligned when we rotate around Z?

• When we run Stream_File_Parser, the poses report an upside-down orientation (180 degrees rotation around X); this is expected.
• This was also true for the previous Pose initialization, which also had Z pointing down; the fact that it is the same is expected, but the fact that either works is not. The whole pose for the previous config was the same in all three axes, which makes sense because it does not have a magnetometer so its initial rotation will be initialized to be about the same.
• When we set the scale factors for the BSP camera pose information to 0, we get no pose changes and --lockRotation doesn't do anything (this is what is expected).
• When we rotate the camera so that it is facing down, the rotation is near [-180, 0, -180]. This does not make sense; it should be 90 degrees off

[russell-taylor]

When we flip the camera-5 configuration file so that it properly describes the upside-down camera configuration, --lockRotation doubles the rotation rather than counteracting it. It looks like we had bugs in both the IMU orientation and --lockRotation that cancelled each other out.

First fix the --lockRotation bug, then add the mounting orientation argument.

[russell-taylor]

When we rotate forwards and backwards, --lockRotation behaves as desired, keeping the same object facing horizontally. Perhaps this is related to the fact that the camera seems to think that it is facing south when it initially determines its orientation (having no compass to lock down rotation around vertical).

[russell-taylor]

Bug fixed in ASDP_Camera_Head that was reporting Eulers with ZYX vs. XYZ order of operations; this fixes the --lockRotation problem.