* see /Vision/Data/kitti_velodyne

* No intensity is provided -> z is used
* No timestamps is provided -> 10Hz is used
* No ring is provided -> elevation angle is computed using HDL 64 
features

[feat] Include external pose info

See merge request !173

* Order is uniformized in all files :
  * Wheel odometer
  * IMU
  * Landmark detector
  * GPS
  * Pose

* Some comments are added/improved

* One validity check was forgotten for each sensor in 
ComputeSensorConstraints

* + logStates validity wasn't checked before using previous poses

* Uniformize the use of CurrentTime
* Uniformize "this" use
* Remove some spaces

* The data can be supplied with time synchronized with network packet
time or with lidar frame header time

* This file must be called calibration_external_sensor.txt, and must be located next to the data file
* Format is 4x4 matrix with space between column elements and line break between line elements

* The multiplication order didn't fit all cases

* This rigidly transforms the trajectory and the maps so the current
pose lays in the synchronized pose measure reference frame

* The search in the external data is optimized for big data files

* A synchronized pose is searched for each point and they are all
represented in Lidar frame

* PC wrapping is adapted to this new feature

* This is very long because the search in the data is not optimized

* External data are used to create a relative transform since last frame 
which is used as a prior pose

* When data are not available a new mode allow to still compute 
ego-motion with constant velocity prior

* Adapt PV wrapping to handle this feature

* Adapt external file parser to handle pose data
* Add classic weight parameters

* Empty shell : needs to compute the constraint

[feat][refact] Keypoints refactoring

See merge request !169

* Enum type cannot be held in an unordered_map

* The keypoint extractor can theoretically be changed in the filter 
property panel. 

* We would like the properties to appear depending on the keypoint types 
that are enabled/disabled

* It is not possible because the two proxies have no communication

* The hints were useless and they are removed

* The user will need to be aware that those parameters have no action on 
the extraction

* Before commit "[feat] Extract all keypoints ", one point over
neighborhood_size was processed to be a potential keypoint, this was
reducing the computation time of keypoints extraction

* Now, all points are processed so the computation time can be increased
in case of dense clouds (32/64/128 Lidar sensors)

* So a parameter is added to downsample the input clouds when extracting
the keypoints to speed up the process.

* The downsampling is random

* /!\ This downsampling only speeds up the keypoints extraction step,
the other steps are mostly driven by the maximum number of extracted keypoints

* If the frame is not 360°, with new space gap the points at the field
of view extremities could be extracted which is not expected. This
problem affects more cases now that first and last points are reenabled.

* Therefore min/max azimuth angles are added to invalidate some points due
to robot/vehicle/pedestrian occlusion or simply because the user wants to
reduce the field of view

* For far objects in empty space, it can be useful to extract
extremities as first points seen (in terms of azimuth) and last points
seen. However, for now, first and last points of the scan lines were
invalidated to solve discontinuity issues.

* This commits reenables all points and allow to check neighborhoods for
first and last points of the line.

* Only space gap is checked for these points because other criteria might
not be valid because of the discontinuity

* No plane points were gotten far away because the lines were long and
the deformation due to Lidar behavior was too strong relatively to the
threshold (2cm)

* A relative threshold is added to fix this problem

* Computations are limited in loops
* Features are not computed if not needed (skip in loop)
* Line fitter is changed from PCA to RANSAC

* Remove saliency parameter

* Add space gap parameter : : minimum number of points that must be 
missed
to consider a feature

* Add space gap parameter : minimum number of points that must be missed
to consider a feature

* Remove saliency parameter

* Angle and intensity based points are filtered to only extract the greatest locally. This means
only one point per line will be extracted on an edge.

* Depth gap is computed without using the lines

* Invalidation step is removed. Points are invalidated using -1 in
values vectors when necessary

* The saliency was really rare used because this kind of points
usually lay in depth gap cases

* This allows to get rid of one parameter for kpts extraction

* For the points which have no line fitted on their left nor right neighborhoods, the depth gap is checked directly on previous left point and next right point

* Criteria to select a line are changed : the width and length of the
line are now analyzed and the angle between each point of the
neighborhood is removed from analysis

* As outputs number is limited, we chose to replace the blobs outputs
(pointclouds + debug arrays) by the intensity edges outputs

* All intensity edges relative parameters are added to PV interface