Xarray-spatial

- Python 3.12 is now supported!
- Python 3.9 & 3.8 are no longer supported.
- Documentation has also been moved to https://xarray-spatial.readthedocs.io/

Pull Requests merged in this release include:
- Add links to docs to readme (800)
- Update readthedocs configuration (798)
- Fix typo in readthedocs config (799)
- Move docs to readthedocs (797)
- Bunch of CI/CD-related fixes (796)
- chore: Remove numpy pin, pin datashader, drop Python 3.7 (789)

The 0.3.7 release is a hot fix for 0.3.6, which has problem with publishing to PyPi
as the package exceeds the limit of 100MB. In this new release, example notebooks are
cleaned up to reduce the package size.

Enhancements
- clear example notebook outputs (786)

With the 0.3.6 release, xarray-spatial now supports python 3.11.
This release focuses on demonstrating the reliability of the library by adding more tests
against GDAL/QGIS.

Enhancements
- Test on Python 3.11 (750)
- Pin numpy version (780)
- zonal stats (764)
- updated citation (769)
- classify.binary: handle NaNs and infinite values (763)
- Test against QGIS, GDAL (744)
- Zonal crosstab user guide notebook enhancement (759)
- cpu_bin with binary search (760)
- outline for housing price feature engineering nb (725)
- More examples to user guide (742)
- Use setuptools (749)
- Update contributor badge (740)
- update pharmacy desserts notebook (723)
- Update README.md (719)
- viewshed gpu notebook (720)
- Fixed feature-proposal.md config (718)

Bug Fixes
- correct docs for proximity (778)
- Zonal_crosstab 3D: Ensure content of input param `values` is preserved (754)
- Multispectral tools: convert data to float32 dtype before doing calculations (755)

The 0.3.5 release mainly addresses the scaling issue in GPU viewshed to gain better accurate triangulation.
The GPU raytraced viewshed should now give comparable results to the CPU version.
However, the 2 versions use 2 different approaches, there can be slightly differences at some points
where a version returns visible while the other considers them as invisible.

Enhancements
- command to get change log (716)
- Added Feature Proposal Template (714)

Bug Fixes
- Improved viewshed rtx. Now result should match the CPU version (715)

The 0.3.4 release primarily a bug fix release but also includes a number of enhancements with a focus on GPU supports.

Enhancements
- NumPy zonal stats: return a data array of calculated stats (685)
- set unit for hotspots output (686)
- More robust cuda and cupy identification (657)
- Remove deprecated tiles module (698)
- Test on python 3.10, remove 3.6 (694)
- moved all tests to github actions (689)
- Add isort to pytest (700)
- Add flake8 to pytest (697)
- Remove unnecessary executable flags (696)
- updated test hotspots gpu (692)
- 3D numpy zonal_crosstab to support more agg methods (687)

Bug Fixes
- Fix rtx viewshed rendering blank image (711)
- Convolve_2d gpu fixes (702)
- focal.mean(): only do data type conversion once (699)
- Update to remote sensing notebook (688)
- focal_stats(): gpu case (709)
- focal apply: drop gpu support (706)
- drop gpu support (705)
- enabled numba.cuda.jit in hotspots cupy (691)

Documentation
- Correct examples in docstrings (703)
- Fix doc build dependencies in CI (683)
- Fix link to Austin road network notebook (695)

- fixed ubuntu version (681)
- Don't calculate angle when not needed (677)
- codecov: ignore all tests at once (674)
- add more tests to focal module (676)
- classify: more tests (675)
- Codecov: disable Numba; ignore tests, experimental, and gpu_rtx (673)
- Improve linter: add isort (672)
- removed stale test files from project root (670)
- User guide fixes (665)
- license year in README to include 2022 (668)
- install dependencies specified in test config (666)
- Pytests for CuPy zonal stats (658)
- add Codecov badge to README
- codecov with github action (663)
- Modernise build system (654)
- classify tools: classify infinite values as nans, natural_breaks: classify all data points when using sub sample (653)
- Add more benchmarks (648)
- Stubbed out function for Analytics module (621)
- Fix doc build failure due to Jinja2 version (651)