Make objects for plate and well acquisitions (#59)

Introduce high content screening plate data structures for Cell Voyager and Molecular Devices systems. The `ConvertToNGFFPlate` converter consumes these plates and writes out ome-zarr NGFF plates. The converter uses dask to stitch the well-images and supports re-chunking and binning.

---------
Co-authored-by: Jan Eglinger <jan.eglinger@fmi.ch>
Co-authored-by: Tim-Oliver Buchholz <tim-oliver.buchholz@fmi.ch>

.pre-commit-config.yaml

@@ -25,7 +25,7 @@ repos:
     rev: 5.12.0
     hooks:
       - id: isort
-        args: ["--profile", "black"]
+        args: ["--profile", "black", --skip=__init__.py, --filter-files]
   - repo: https://github.com/psf/black
     rev: 23.9.1
     hooks:

examples/CellVoyagerStackAcquisitionExample.py

@@ -0,0 +1,50 @@
+import shutil
+from pathlib import Path
+
+from faim_hcs.hcs.acquisition import TileAlignmentOptions
+from faim_hcs.hcs.cellvoyager import StackAcquisition
+from faim_hcs.hcs.converter import ConvertToNGFFPlate, NGFFPlate
+from faim_hcs.hcs.plate import PlateLayout
+from faim_hcs.stitching import stitching_utils
+
+
+def main():
+    # Remove existing zarr.
+    shutil.rmtree("cv-stack.zarr", ignore_errors=True)
+
+    # Parse CV plate acquisition.
+    plate = StackAcquisition(
+        acquisition_dir=Path(__file__).parent.parent
+        / "resources"
+        / "CV8000"
+        / "CV8000-Minimal-DataSet-2C-3W-4S-FP2-stack_20230918_135839"
+        / "CV8000-Minimal-DataSet-2C-3W-4S-FP2-stack",
+        alignment=TileAlignmentOptions.GRID,
+    )
+
+    # Create converter.
+    converter = ConvertToNGFFPlate(
+        ngff_plate=NGFFPlate(
+            root_dir=".",
+            name="cv-stack",
+            layout=PlateLayout.I384,
+            order_name="order",
+            barcode="barcode",
+        ),
+        yx_binning=2,
+        dask_chunk_size_factor=2,
+        warp_func=stitching_utils.translate_tiles_2d,
+        fuse_func=stitching_utils.fuse_mean,
+    )
+
+    # Run conversion.
+    converter.run(
+        plate_acquisition=plate,
+        well_sub_group="0",
+        chunks=(2, 1000, 1000),
+        max_layer=2,
+    )
+
+
+if __name__ == "__main__":
+    main()