Asf-tools

Removed
- Support for Python 3.8 and 3.9 has been removed.

Fixed
- Changed relative link to the watermasking readme in the repo readme to the full URL, so that the link is valid when readme content is mirrored in hyp3-docs

Added
- A description of the `asf_tools.watermasking` sub-package has been added to the [`asf_tools` README](src/asf_tools/README.md)
- Installation instructions for `osmium-tool` have been added to the [`asf_tools.watermasking` README](src/asf_tools/watermasking/README.md)

Fixed
- `osmium-tool` dependency handling. Because `osmium-tool` is not distributed on PyPI and thus is not installed when `pip` installing `asf_tools`, `asf_tools` will now raise an `ImportError` when `osmium-tool` is missing that provides installation instructions. Note: `osmium-tool` is distributed on conda-forge and will be included when conda installing `asf_tools`.

Added
* Scripts and entrypoints for generating our global watermasking dataset added to `watermasking`.

Added
* You can choose whether the `ts` (threat score; default) or `fmi` (Fowlkes-Mallows index) minimization metric is used for the flood mapping iterative estimator:
* the `flood_map` console script entrypoint now accepts a `--minimization-metric` argument
* the `asf_tools.hydrosar.floopd_map.make_flood_map` function now accepts a `minimization_metric` keyword argument
* The flood mapping iterative estimator will ignore waterbodies smaller than a minimum number of pixels (default = 0)
* the `flood_map` console script entrypoint now accepts a `--iterative-min-size` argument
* the `asf_tools.hydrosar.floopd_map.make_flood_map` function now accepts a `iterative_min_size` keyword argument

Changed
* The HydroSAR code (`flood_map`, `water_map`, and `hand`) in `asf_tools` has been isolated to an `asf_tools.hydrosar` sub-package
* The `asf_tools.hydrosar.flood_map.iterative` estimator now runs with a maximum step size of 3 instead of the default 0.5.
* The `asf_tools.hydrosar.flood_map.iterative` estimator now uses the mean of the iterative bounds at the initial guess.
* the known water threshold used to determine perennial water when creating flood maps will be calculated `asf_tools.hydrosar.flood_map.get_pw_threshold` if not provided
* `get_epsg_code` and `epsg_to_wkt` have been moved from`asf_tools.composite` to `asf_tools.util`
* `read_as_array` and `write_cog` have been moved from`asf_tools.composite` to `asf_tools.raster`
* `get_coordinates` has been moved from`asf_tools.flood_map` to `asf_tools.util`

Deprecated
* The `asf_tools.hydrosar` sub-package is being moved to the [HydroSAR project repository](https://github.com/fjmeyer/hydrosar) and will be provided in a new pip/conda installable package `hydrosar`. The `asf_tools.hydrosar` subpackage will be removed in a future release.

Fixed
* Reverted the special handling of nan values introduced in v0.5.2, now that GDAL v3.7.0 has been released.

Added
* Updated tests to use a water_map/flood_map job that was created using a known input SLC

Fixed
* Patched issue with gdalcompare.py's handling of nan values by allowing one differences between two rasters
that contain nan values. This patch can be remove once the upstream fix is released within GDAL (likely v3.7.0)
* Fixed incorrect datatype being set for `flood_mask` GeoTIFFs leading to missing nodata.