Fractal-tasks-core

* Tasks:
* Add percentile-based rescaling to calculate registration task to make it more robust (\848)
* Dependencies:
* Relax pandas constraint to `<2`.
* Relax torch constraint to `<=3.0.0`.
* Relax numpy constraint to `<2.1.0`.
* Add python 3.12 to the CI matrix (\770).
* Chores:
* fix future warning when using Series.__getitem__ with positional arguments.

* Testing
* Use latest version of Zenodo tiny-ome-zarr dataset (\817).
* Relax pip-version constraint in external-packages-manifest tests (\825).
* Run pip-based CI also regularly and on-demand (\829).
* Update GitHub actions for upload/download/coverage (\832).
* Dependencies:
* Require `pydantic<=2.8.2` (\836).
* Require `stackview<=0.9.0` (\829).
* Documentation:
* Bump `mkdocstrings` and `mkdocstrings-python` to support `griffe` v1 (\818).

* Tasks:
* `image_glob_patterns` are renamed to `include_glob_patterns` in Convert Cellvoyager to OME-Zarr (regular & multiplexing) (\812).
* Convert Cellvoyager to OME-Zarr (regular & multiplexing) gain exclusion patterns to exclude specific patterns of images from being processed (\812).
* Fix issue with arbitrary acquisition names in Convert Cellvoyager Multiplexing to OME-Zarr (\812).
* In Convert Cellvoyager to OME-Zarr (regular & multiplexing), handle channels in the mrf metadata file that aren't present in the mlf metadata better (\812).
* In Convert Cellvoyager to OME-Zarr, improve plate metadata for image list when multiple plates with the same plate name are processed (\812).
* Catch errors for missing mlf & mrf files better in Convert Cellvoyager to OME-Zarr (regular & multiplexing) (\812).
* Drop defusexml dependency for cellvoyager metadata conversion (\812).
* Rename `Maximum Intensity Projection HCS Plate` task to `Project Image (HCS Plate)` (\814).
* Expose selection of projection modes in `Project Image (HCS Plate)`: MIP, MINIP, MEANIP & SUMIP (\814).
* Rename task function from `maximum_intensity_projection` to `projection` and modified parameters in `fractal_tasks_core.tasks.io_models.InitArgsMIP` (\814).

* Core-library
* Add `create_roi_table_from_df_list` library function in `fractal_tasks_core.v1.roi`: It combines a list of ROI table dataframes into an AnnData ROI table and handles repeating labels (\811).
* Move `create_well_acquisition_dict` & `_split_well_path_image_path` from `fractal_tasks_core.tasks._registration_utils.py` & `fractal_tasks_core.tasks._zarr_utils` into `fractal_tasks_core.utils` (\811).
* Tasks:
* Fixes issue 810 for Cellpose task: Avoids creating duplicated entries in ROI tables when masking ROI table input was used (\811).

* Core-library and tasks:
* Switch all core models to Pydantic V2 (\793).
* JSON Schema generation tools:
* Move JSON-Schema tools to Pydantic V2 (\793).
* Testing:
* Remove dependency on `pytest-pretty` (\793).
* Update `manifest_external_packages.yml` GitHub Action so that it installs the current `fractal-tasks-core` (\793).

* Tasks:
* Fix issue with masked ROI & relabeling in Cellpose task (\786).
* Fix issue with masking ROI label types in `masked_loading_wrapper` for Cellpose task (\786).
* Enable workaround to support yx images in Cellpose task (\789).
* Fix error handling in `calculate_registration_image_based` (\799).
* Fix minor issues with call-signature and type hints in `calculate_registration_image_based` (\799).