Improved depth estimation of mono features

[matlabbe]

Added parameters:

• OdomF2M/BundleUpdateFeatureMapOnAllFrames: Update 3D local feature map on every frame with bundle adjustment. Recommended if Vis/DepthAsMask=false and Mem/UseOdomFeatures=true so that features without depth are better triangulated on every frame (not only on keyframes). If disabled, the feature map is updated only when a new keyframe is added (legacy approach).
• OdomF2M/FloorThreshold: Only track features in 3D feature map that are over this threshold (height in base frame). Can be useful to ignore reflections on the floor. 0 means disabled.
• OdomF2M/InitDepthFactor: Depth factor used to initialize depth of features without depth. Depth = Factor * fx.

TODO: Re-run vslam benchmark datasets with and without OdomF2M/BundleUpdateFeatureMapOnAllFrames to see if it doesn't affect too negatively the results. Ideally, we would want OdomF2M/BundleUpdateFeatureMapOnAllFrames to be true by default so that depth of mono features can be correctly estimated all the time.

[matlabbe]

Euroc datasets are the more negatively impacted by OdomF2M/BundleUpdateFeatureMapOnAllFrames=true, we should keep it to false by default.

[matlabbe]

I tested again Euroc with live visualization and the results at 1 Hz are pretty close to original results. It seems that a not so accurate loop closure "sometimes" happens, which add significant error, though if I remove that link, the results are close to paper.

I'll still keep OdomF2M/BundleUpdateFeatureMapOnAllFrames=false by default till I have more real experiments with OdomF2M/BundleUpdateFeatureMapOnAllFrames=true.