Before running PCA, check that points are roughly aligned by:
 - checking that each point is not too far from the approximate line
 - checking that the angle between successive points and the approximate line is not too big
After PCA, the linearity feature is computed and checked.

This produces better results in less processing time.

[feat] Add invalidation parameter for oblique surfaces

See merge request !115

* We encountered some environments requiring the use of very oblique
surfaces

* A member MinBeamSurfaceAngle is added to the SSKE to allow the use of
such surfaces

* Update wrappings consequently

* Default value is 20°

[doc] Update README and LidarView tutorial

See merge request !113

This file is displayed in LidarView's help.
It has been auto-generated from How_to_SLAM_with_LidarView.md by https://www.browserling.com/tools/markdown-to-html

[fix] Fix bugs and rename variables

See merge request !112

* Invalidation was missing some points

* After gathering lidar residuals and external sensor residuals, the 
names were not adapted in the local optimizer and could be confusing.

To be able to fit a line, the neighborhood must contain at least 2 
points

[fix] Correct debug array error

See merge request !109

Allow compilation on Win10 MSCV 2015

See merge request !107

Release v1.4

See merge request !111

Fix refactoring bugs

See merge request !108

The last important refactorings (!103 and !101) brought some bugs.

- Lidar ICP residuals were added twice to the ceres problem, resulting in much slower optimization.

- Set the max number of threads to use for parallel omp for loops to the number of keypoints types
instead of a hard-coded "3 threads".

- Do not build blob KDTree during ego motion registration step

- Remove unsused PoseArray member in KeypointsMatcher

- Reset all keypoints clouds even if unused.
  This is necessary otherwise old blob cloud is still displayed even if blobs are disabled.

[feat] Split Matching and optimization

See merge request !101

* Lidar and external sensor residuals are gathered in the same vector
* This vector is refilled entirely at each ICP step

* Split KeypointsRegistration file into LocalOptimizer and 
KeypointsMatcher files.

* LocalOptimizer

   * It is a member of the SLAM class.
   * A residual structure is created in this file. It contains pointers 
to a cost function and a loss function (optimization estimator to remove 
outliers).
   * It receives the ceres LiDAR residuals.
   * It will be able to receive other sensors' residuals.
   * It is entirely cleared at each registration step.
   * It contains a ceres problem pointer (the destructor is updated). 
This problem is created and filled after each new matching step of ICP. 
The problem is conserved in order to be able to recover covariance from 
the optimization after convergence for last ICP iteration.

* KeypointsMatcher

   * It is an object created locally in Ego-Motion and/or Localization 
functions.
   * It builds the LiDAR residuals (as described above) from matching 
between frame keypoints and map neighborhoods.
   * It contains a vector of residuals (directly added in the matching 
result structure).
   * A residual is also added to the matching info structure.

Optional calibration in PV wrapping

See merge request !106

Calibration is only needed to reorder laser ids.
However, we don't really care if these laser rings are not properly sorted,
as SSKE consider each ring independently from other ones.
To simplify user interface, this second entry is hidden from selection dialog.
It can still be accessed later by manually changing inputs.

If calibration input is empty (such as for HDL64),
or if the vertical angle array isn't available, ignore calibration.

This fixes using HDL64 data with PV wrapping, where calibration is empty.

[refact] Gather all keypoint types relative structures and functions

See merge request !103

* The lambda functions and omp sections are replaced by an omp "for"
loop
* Other redundant processes are replaced by loops

* UseKeypoints is a booleans std map which allows to select which
keypoints to use in the optimization
* KeypointTypeNames is a strings std map which contains the respective
strings used in printing processes
* Remove FastSlam
* Update wrappings to adapt to new UseKeypoints map
* FastSlam is replace by UseBlobs
* Getters and setters are updated to be relative to UseKeypoints map

* "edges points/keypoints/weights/matches"  -> "edge ..."
* "planes points/keypoints/weights/matches" -> "plane ..."
* "blobs points/keypoints/weights/matches"  -> "blob ..."
* "planars" -> "planes"
* + Alignment

All functions and features linked to keypoints exist for edges, planes
and blobs.
This leads to triplicated parts of code and make the whole process
hardly adjustable (add or remove keypoint types). This MR intends to fix
part of these issues and to uniformize the way to access a keypoint
relative feature with EDGE / PLANE / BLOB key words.

* Such features are stored in maps which pair the keypoint type with the
feature.
* The relative functions are adapted to take the keypoint type as
argument.
* When possible, a loop processes iteratively each keypoint type.
* To be able to loop on keypoint types, a list of keypoint types is
added in enums header file.

It is to note that all parts couldn't be replaced by loops because of
some specific checks (e.g. are blobs used?) or some specific output
features (e.g string in cout, UI relative names in wrappings). These
tasks should still be changed by hand if some keypoint types is
modified.

Fix catch-by-value warning and fix undistortion time range

See merge request !105

- We were currently returning the reverse mapping.
  This reverses the mapping by an additional SortIdx call.
- Change default behavior of Utils::SortIdx to sort by default in ascending order, like std::sort.

Auto compute angular resolution for SSKE

See merge request !102

This parameter will now be auto-computed with first frame.
Rename AngleResolution to AzimuthalResolution.

The horizontal resolution is computed for the first frame only.
This process still requires a scan sorted by laser rings,
each ring being sorted by increasing azimuth/timestamp.

NLasers -> NbLaserRings
pclCurrentFrame -> Scan
pclCurrentFrameByScan -> ScanLines

[feat] Change keypoint invalidation criteria

See merge request !93