Pyinterpolate

* (setup) added `pyogrio` to dependencies due to the new `fiona` version (1.8.22) and `gdal` errors.

2022-10-21
----------

* (debug) control of data *dtypes* after transformations and preparation of `PointSupport` and `Blocks`,
* (debug) updated data selection methods for Poisson Kriging to avoid mixing column of numerical and non-numerical values in a single numpy array, (it makes algorithm faster),
* (update) updated tutorials,
* (feature) check area and point support indexes with `smooth_area_to_point_pk()`,
* (docs) updated docstrings for `calculate_covariance()` and `calculate_semivariance()` functions.


2022-10-18
----------

* Semivariogram `Deconvolution` takes possible model types as a parameter,
* Semivariogram `Deconvolution` uses **basic** set of variogram models (*spherical*, *linear*, *power*, *exponential*),
* New class: `DirectionalVariogram` calculates experimental variograms in four directions along with isotropic variogram,
* Corrected directions (angles were described wrong 0 degrees is W-E, -90 deg is N-S direction),
* Directional variogram calculations are faster due to the change of selection method (only non-weighted case in this release),
* Numpy's `sqrt()` method casts ints into floats (see `Issue 306 <https://github.com/DataverseLabs/pyinterpolate/issues/306>`_),
* Users can pass a nugget for `autofit()` method of `TheoreticalVariogram()` class, the same for `Deconvolution()` process.

2022-10-08
----------

* new test dataset with regular blocks,
* more tests for `Deconvolution`, `area_to_point_pk()`, `area_to_area_pk()`, and `centroid_based_pk()`,
* if there are no values for a given lag in experimental variogram `RunetimeError()` is raised,
* `average_block_to_block_semivariances()` appends 0 to the lags with 0 points,
* `calculate_block_to_block_semivariance()` - valid calculation of number of point pairs.


2022-09-29
----------

* experimental variogram, covariogram, and variogram cloud function and classes check if there are NaN's in the input data and raise `ValueError`,
* the length of major and minor axes of a directional variogram ellipsis are calculated differently from the `tolerance` parameter, (now we have a less of chaos),
* tutorial for directional variograms (Basic),
* updated `download_air_quality_poland()` function, now it can store downloaded data,
* updated documentation.

2022-09-04
----------

* module `io_ops` renamed to `io`,
* the refactored function `read_point_data` (old) into `read_txt`, new functions to read csv and blocks data,
* the new objects to store block data and its point-support: `Blocks` and `PointSupport`,
* Kriging is now supported by **functions**, not by classes, to speed up some calculations. In the future, classes will be introduced again,
* user has much more control over the variograms development. `ExperimentalVariogram` class calculates *variance*, *covariance*, and *semivariance*, has own plotting function. `TheoreticalVariogram` has more models to fit, and gives more control to search for the best fit - the algorithm searches over ranges and sills. Nugget is still fixed,
* module `pipelines` has the function for the block data smoothing (area-to-point Poisson Kriging), the class for block data filtering (area-to-area Poisson Kriging), the kriging comparison class, and method to download sample air pollution data,
* there are many small changes and API transformations... The package is faster and more stable,
* it works with Python 3.7, 3.8, 3.9, and 3.10,
* Ordinary and Simple Kriging of large datasets may be performed in parallel,
* the package has a few warnings and raises custom errors,
* `setup.py` is removed, now package installs from `setup.cfg`,
* data structures are more complex, but they allow user to be more flexible with an input.


2021-12-31
----------