Spectrochempy

Pin pint and numpy version for compatibility
Fix issue 758 : serialization error for tuple containing a ndarray

New features
--------------
* Compatibility with Python 3.12
* Add the possibility to pass a colormap normalization to the `plot` method.
* Add the possibility to use several sets of experimental conditions
in `ActionMassKinetics` class.
* Add the possibility to read Thermo high speed series files
* Add Stejskal-Tanner kernel for 2D IRIS
* Fancy indexing using location now supported.
* Add an example for NMR processing of a series of CP-MAS spectra.
* Add an example for processing NMR relaxation data
* Add an option to `read_topspin` to create ``y`` coordinates
of pseudo-2D NMR spectra from a file (e.g. ``vdlist`` ).
* Add option to plot to add markers on curves
* Add a new method to the `Optimize` class to perform a least-square fitting. It is
based on the `scipy.optimize.least_squares` function, allowing much faster operation
for simple curve fitting
* Add the possibility to define user-defined functions in the `Optimize` class.
* Traceback are now fully displayed when an error occurs in a script.

Bug fixes
----------
* Sorting coordinates now work with multi-coordinates axis.
* Fix a bug when concatenating datasets with multi-coordinates axis.
* Fix a bug in coordset definition for integration methods.
* Fix coordinates definitions in Analysis methods.
* Fix a bug in `write_csv` when the filename was provided as a string (issue 706)
* Fix issue 716
* Fix issue 714 : show versions of dependencies now working

Breaking changes
------------------
* Changed the default QP solver (quadprog -> osqp): The new solver is compatible with
python 3.11 and later. Fastness and robustness are improved. The quadprog solver can still be
used if available
* Change the default value of the `whiten` parameter in the `FastICA` class to
`unit-variance` instead of `arbitrary-variance` for compatibility with ScikitLearn
1.3 and later
* Colormap normalization for `surface`, `image` and `map` plot methods has been
changed for consistency with matplotlib default. The former behaviour can be obtained
by passing a `norm` parameter to the `plot` method (see userguide/plotting).

New features

* A new reader has been added: `read_wire` (alias `read_wdf``) to read data from
the .wdf format (WDF) files produced by the ReniShaw WiRe software.
This reader is based on the `py_wdf_reader <https://github.com/alchem0x2A/py-wdf-reader>`_ package.
* Added an example for NMR processing.
* Peak finding now handles dimensions other than x (important when data are transposed, or when working on a slice in dimension other than x)
* Analysis methods now accept complex input data, but work only on the real part of the data.

Bug fixes

* Fix a bug when slicing dataset with an array or list of index: Multi-coordinates
were not correctly handled.
* Increase the value of the coordinate linearization condition from 0.1% to 1% spacing variation.
(linearization was sometimes lost when slicing)
* Fix a missing correction for non-negative spectra in MCR-ALS.

New features

* `~spectrochempy.plot_multiple` method now accept keyword arguments to change the default
plot style of the different spectra.
* `BaselineCorrection` class has been renamed into
`~spectrochempy.Baseline`, and there are changes in the way it
is now used. It allows to perform baseline correction
on a dataset with multiple algorithms. `BaselineCorrection` is still valid but deprecated.
* Three new baseline algorithms have been added to the new Baseline processor:
`~spectrochempy.rubberband`, `~spectrochempy.asls` and `~spectrochempy.snip` .
* Filters has been refactored. A new `Filter` processor class allows to define various
filters and apply them to a dataset. See `Filtering and Smoothing` tutorials and `Filter`
for details. Note: Backward compatibility is ensured with the previous `smooth` and `savgol_filter` methods.
* A `whittaker` filter has been added to the `~spectrochempy.Filter` processor class. See `Filtering and Smoothing`
tutorials and `~spectrochempy.Filter` for details. A method is also available: `~spectrochempy.whittaker`.
* A `~spectrochempy.denoise` method based on PCA analysis has been added which allows to apply a denoising filter to a 2D dataset.
* A `~spectrochempy.despike` method has been added to the `~spectrochempy.Filter` processor class.
It allows to remove spikes from a 1D or 2D dataset. This close issues 688.
* New examples has been added to the gallery

Bug fixes

* Docs problems fixed (687).

Deprecations

* `parameters` method of Analysis configurables is now deprecated in favor of `params`.
* The `BaselineCorrection` processor has been deprecated in favor of `~spectrochempy.Baseline` .
* `abc` (and its alias `ab`) method has been deprecated in favor of `~spectrochempy.basc`.
* `savgol_filter` method has been deprecated in favor of `~spectrochempy.Filter` processor class or the `~spectrochempy.savgol` method.

Bug fixes

* issue 664: wrongly defined xaxis for opus file.
* issue 668: fix bug with fancy indexing of datasets (fix also 663).
Both issues were related to the linearization of the coordinates,
which was not done correctly in some cases. This is now fixed.