Running `test_skeleton_fitter.py` as a script will still show the plots,
but for convenience of running the test suite from the command line,
they are not shown anymore when executed with `pytest`.

`pytest` needs to be executed from `tests/`, since the tests all
assume paths relative to this directory.

This makes them easier to copy-paste into the terminal or an editor.

Most MarkDown renderers only recognise paragraphs if they are
separated by an empty line.

The added how-to was based on a clean installation on a new machine.

The version ranges indicated previously were too broad, making it difficult
for pip to resolve dependency conflicts. The relatively narrow version ranges
indicated here were tested to work.

Remove image.py and all dependencies => stopped loading of all images when building imagesequences

See merge request cribe001/dlc_flytracker!11

+ Added show plot for gen_dlt_coefs_from_paths
+ Other small changes

+ added first_frame and missing frames in BatchSettings
+ fixes issues with np.arrays (dtype=object)

Added posibility to know the id of the calibration point when plotted during error analysis, better allow identification of points to remove

+ Made finding shared substring (for image files) much faster
+ some bug fixing

Major development of the various modules + many bugs were solved

See merge request cribe001/dlc_flytracker!10

Added intrinsic_from_extrinsic_and_coordinates and extrinsic_from_intrinsic_and_coordinates in camera.calib
+ renamed coords to coord when referring to a single set of 3d coordinates [x, y, z]

Made nb_dim a optional input for estim_dlt_coef

+ added example _full.pickle
+ removed Jupyter notebooks

+ Add main() in test functions
+ Changed res_method to fit_method

+ Merged camera.py and virtual_camera.py into objects.py

Wrote a test function test_calib_multi_focus to test calibration of one camera for multiple focus distances and focal lentghs

Added estimMDLT to estimate modified DLT coefficient (higher accuracy and orthogonal matrix -> needed to estimate quat)
+ Various changes to calib.py and debugging of related test functions

+ Modified settings to match recordings
+ Added sorting of frame numbers in ImageSequence class
+ Corrected back get_rotation_angle_btw_vectors
+ Some correction of docstrings

+ remove TODOs about using [0., 0., 1.] instead of [1., 0., 0.], was not necessary

+ Minor edits

+ Updated printed message during analysis
+ Changed folder tree + renamed some parameters files
+ Minor edits in Readme.md

Add correction factor to convert units of 3d coordinates use for the calibration (eg millimeters to meters)

+ Added functions to update .yaml with wing (skeleton/geometry) 3d coordinates from a .csv file that contains 2d coordinates

Minor edits (TODOs, docstrings, renaming variables, ...) and some minor debugging (dealing with nans, ...)

- DLT coefficient to/from intrinsic and extrinsic camera parameters (pinhole model)
- Improve modeling of 3d object, camera, mirror on a gimbal, and virtual camera
- test rotation matrices and pinhole camera model

+ Changed convention for normal vector of non-rotated object (from [1, 0, 0] to [0, 0, 1])
+ Use extrinsic rotation 'xyz' instead of intrinsic 'ZYX' for scipy.transform.Rotation
+ Implemented quaternions for calibration, Object3D, Camer, etc. But still need to implement them in skeleton_fitter
+ Improve documentation (doctstrings)