Gstlearn

=============
- Gneiting covariance function has been added (with its test)
- New SpaceComposite class added (with its test)
- A method SpacePopint::spacePointOnSubspace has been added to project a
point on a sub-space of a SpaceComposite (there is still Memory copies!)
- A sample index (in the Db) and a boolean (target) have been added to the SpacePoint members
- Adding the BASH version for files "compare_test" and "update_test" used to handle non-regression output files.
- Re-export acknoledge_gstlearn function in packages

Miscellaneous
-------------
- No more Leak in non-regression tests for Data
- Add the location of non-regression specific files if needed in compare_test and update_test

=============
- Fix Zenodo configuration to restore DOI generation

=============
- AMatrix: Correction in linearCombination (when processing Symmetric matrices and if this coincides with mat*)
- Seed has been removed for arguments in all SPDE functions. Use law_set_random_seed before a call instead).
- Bug with K-Bessel function (problem for large values) has been fixed (by using std::cyl_bessel_k for Windows and Linux and a trick for MacOS for which std::cyl_bessel_k is not available).
- The interfaces to deal with non stationarity of covariances parameters (angles, ranges, sills, ...) have been totally changed. See Tuto_NoStat.ipynb for a full description.
- SPDE now available in multivariate context (see Tuto_multivarSPDE.ipynb)
- Improve R package stability
- Reduce drastically moving neighborhood computation time
- Correcting severe bug in quick interpolators (movingAverage, inverseDistance, ...)
- Fix a bug in AMatrix::resetFromVVD (byCol logic was inverted)
- law_exp_sample() now returns a VectorDouble (instead of double *)
- In Db::createFilLRandom, use of 'ncode' has been corrected
- Method computeLogLike() is renamed in computeLogLikelihood()
- Matrix methods named *Norm* use pointers to Matrix rather than references
- New method to initialize AStringable(const String& title) which prints the title in display()
- Internal function mesArg() is abandoned in favor of checkArg()
- A new bench_NF to help optimizing the read and write in Neutral Files
- Adding method 'display' for any item of class VectorT
- Add a DOI for gstlearn software published by Zenodo
- "How to cite" section in READMEs updated
- New BibTeX file format
- Covariance type are modified: BESSEL_K into MATERN and BESSEL_J into BESSELJ
- The constructor of MeshETurbo() defines rotation using 'angles' rather than 'rotmat'
- New class SpatialIndices and corresponding methods (CGI, LIC, GIC, ...) as in RGeostats (micro not yet available).
- Names of variables used to compute the Variogram are stored in the Vario object
- Empirical anamorphosis: plugging Normal score; parameterizing the dilution
- Adding the file dev/Valgrind_Statistics for a follow up of software performances.

=============
- Fix Windows package for python 3.10 (Numpy compatibility)

=============
- Fix getColormap in plot.py

=============
- migrateAllVariables() is now a method of DbGrid (not a static) with additional arguments
- R 4.4.1 support
- Modify 'flag_cell' into 'flagCell'
- Modify 'flag_integer' into 'flagInteger'
- Modify flag_add_rank into flagAddSampleRank
- change 'flag_add_coordinates' into 'flagAddCoordinates' in DbGrid::createSubGrid() and DbGrid::reset()
- Increase the size of the Enums from 32 to 40.
- Model::evalDriftMat() is replaced by Model::evalDriftMatrix()
- Model::covMatrixMS() is replaced by evalCovMatrixSymmetric()
- Model::covMatrixM() is replaced by evalCovMatrix()
- Model::covMatrixV() is replaced by evalCovMatrixV() (option added)
- Db::getActiveSampleNumber() (deprecated) is replaced by getSampleNumber(true)
- Use Eigen library for content storage of all dense matrices