Ravenpy

--------------------

* Set base required `geopandas` to v1.0. (PR 394)
* Removed the pin on `pyogrio` (set by `geopandas` now). (PR 394)
* Removed the `requests` dependency (now using `urllib`/`urllib3`). (PR 394)

Internal changes
^^^^^^^^^^^^^^^^
* The cookiecutter template has been updated to the latest commit: (PR 386)
* `ravenpy` now uses a `src`-layout for the package.
* `HISTORY.rst` has been renamed to `CHANGELOG.rst`.
* `ruff` checks have replaced most of the `flake8` checks.
* `ravenpy` now has a `CODE_OF_CONDUCT.md` file.
* Many `numpydoc`-style docstrings have been adjusted for consistency.
* Added `setuptools` to the `gis` build recipe to ensure that the `gdal` bindings are built successfully. (PR 400)
* Modified the sub-basin and channel profile extraction functions to correctly set the river length to zero and set default values for reach attributes in sub-basins with no channel routing (i.e., sub-basins with lakes or headwater basins). (issue 354, PR 401)
* Improved the HBV-EC emulator by adding parameter information (name, definition, and Raven default values), fixed the variable name for the adiabatic temperature lapse rate, and added an alias for rain snow fraction to match other emulators. (PR 404 and 408)
* Modified the `sphinx` configuration to better support SVG and to remove incompatible elements from the PDF build. (PR 407)

--------------------

* Pinned `pint` below version 0.24 due to a breaking change in their API. (PR 375)
* Pinned `numpy` below v2.0.0 due to a breaking change in their API. (PR 378)
* Update `raven-hydro` to v0.3.1 and `RavenHydroFramework` to v3.8.1. (PR 378)
* Fixed bug in `Config.duplicate` dating from the switch to Pydantic V2 in 0.13 (PR 367)

Internal changes
^^^^^^^^^^^^^^^^
* Synchronize several dependencies between `pyproject.toml`, `environment*.yml`, and `tox.ini`. (PR 378)
* Drop the code formatting conventions for Python3.8, extend to Python3.11 and Python3.12. (PR 378)
* Addresses a bunch of small warnings in the pytest output. (PR 378)

--------------------

* Upgraded `owslib` to `>=0.29.1`. (PR 358)
* All operations that open NetCDF files or DAP links accept an `engine` argument. The default for all of these is `h5netcdf`. (PR 358)
* Added `pydap` as an alternate backend for opening DAP links. (PR 358)
* Fixed buggy CustomOutput command. (PR 360)
* Make sure config and output paths are absolute. (PR 360)

Internal changes
^^^^^^^^^^^^^^^^
* Added some development dependencies that were missing to the `environment.yml`. (PR 358)
* `test_climpred_hindcast_verif` is now skipped for Python3.10 builds. It seems to only fail on the particular version of Python. When examining the dependencies, other than the Python version (and ABI version), there are no differences in the environments between Python3.10 and Python3.11. Possibly an issue with `climpred`. (PR 358)
* Temporarily disabled tests for macOS on GitHub due to architecture changes. (PR 358)
* Pinned `pyogrio` below v0.8.0 until `geopandas` supports it. (PR 363)
* Updated linting dependencies to the latest versions. (PR 363)

--------------------

* Add support for new processes and methods added in Raven v3.8. (PR 335)
* Add Interpolation command options. (PR 338)
* Let VegetationClass records contain symbolic expressions. (PR 338)
* Add support for custom RV subclasses. (PR 338)
* Use HRU_ID (if available) instead of SubId in BasinMaker reservoirs extraction logic. (PR 338)
* Added support for Python 3.12 and dropped support for Python3.8. (PR 341, PR 343)
* Added support for `raven-hydro` v0.3.0 and `RavenHydroFramework` to v3.8. (PR 341, PR 351)
* `ravenpy` now requires `xclim` >= v0.48.2, `xarray` >= v2023.11.0, and `pandas` >= 2.2.0. (PR 341)
* Now automatically filters HRUs based on the ``hru_type``. (issue 340, PR 334)

Internal changes
^^^^^^^^^^^^^^^^
* Updated GitHub publishing workflows to use Trusted Publisher for TestPyPI/PyPI releases. (PR 341)
* Added Dependabot to keep dependencies up-to-date. (PR 342)
* Now using step-security/harden-runner action to harden GitHub Actions runners. (PR 341)
* Adjusted GitHub Workflows to test against Python 3.9, 3.10, 3.11, and 3.12. (PR 341, PR 343)
* Updated the build-system requirements when testing with `tox` to use newer `setuptools` and `wheel` versions when building `gdal`. (PR 341)

--------------------

* Fixed problem with scalar elevation in netCDF files parsed with `nc_specs`. (issue 279, PR 323)
* Added notebook on sensitivity analysis. (PR 320)
* Updated Notebooks 03 and 04. (PR 319)
* Upgrade to `pydantic` v2.0. (PR 326)
* Pin `cf-xarray` for Python3.8. (PR 325)
* Fix `Coveralls` Workflows. (PR 328)
* Fix notebook execution. (PR 329)
* Refactor and simplify testing data fetching. (PR 332)

Breaking changes
^^^^^^^^^^^^^^^^
* Update to `pydantic` v2.0. (PR 326)
* Added `h5netcdf` as a core dependency to provide a stabler backend for `xarray.open_dataset`. (PR 332)
* Switched from `autodoc_pydantic` to `autodoc-pydantic` for `pydantic` v2.0+ support in documentation. (PR 326)

Internal changes
^^^^^^^^^^^^^^^^
* Removed some redundant `pytest` fixtures for running `emulators` tests.
* `"session"`-scoped `pytest` fixtures used for hindcasting/forecasting are now always yielded and copied to new objects within tests.

--------------------

* `RavenPy` now uses `platformdirs` to write `raven_testing` to the user's cache directory. Dynamic paths are now used to cache data dependent on the user's operating system. Developers can now safely delete the `.raven_testing_data` folder in their home directory without affecting the functionality of `RavenPy`.
* Updated `raven-hydro` to v0.2.4 to address CMake build issues.

Breaking changes
^^^^^^^^^^^^^^^^
* In tests, set `xclim`'s missing value option to ``skip``. As of `xclim` v0.45, missing value checks are applied to the ``fit`` indicator, meaning that parameters will be set to `None` if missing values are found in the fitted time series. Wrap calls to ``fit`` with ``xclim.set_options(check_missing="skip")`` to reproduce the previous behavior of xclim.
* The `_determine_upstream_ids` function under `ravenpy.utilities.geoserver` has been removed as it was a duplicate of `ravenpy.utilities.geo.determine_upstream_ids`. The latter function is now used in its place.

Internal changes
^^^^^^^^^^^^^^^^
* Added a GitHub Actions workflow to remove obsolete GitHub Workflow cache files.
* `RavenPy` now accepts a `RAVENPY_THREDDS_URL` for setting the URL globally to the THREDDS-hosted climate data service. Defaults to `https://pavics.ouranos.ca/twitcher/ows/proxy/thredds`.
* `RavenPy` processes and tests that depend on remote GeoServer calls now allow for optional server URL and file location targets. The server URL can be set globally with the following environment variable:
* `RAVENPY_GEOSERVER_URL`: URL to the GeoServer-hosted vector/raster data. Defaults to `https://pavics.ouranos.ca/geoserver`. This environment variable was previously called `GEO_URL` but was renamed to narrow its scope to `RavenPy`.
* `GEO_URL` is still supported for backward compatibility but may eventually be removed in a future release.
* `RavenPy` has temporarily pinned `xarray` below v2023.9.0 due to incompatibilities with `xclim` v0.45.0`.