[ds-api] Modification of dataset with float-type layer results in JSON dump error (#358)

* Create failing test

* Fix JSON serialization error for numpy float

* applied changes: all properties_floating_type_to_python_type values are strings

Author: erjel

tests/test_dataset.py

@@ -196,6 +196,21 @@ def test_open_dataset() -> None:
 
     assert len(ds.properties.data_layers) == 1
     assert len(ds.properties.data_layers["color"].wkw_magnifications) == 1
+    
+    
+def test_modify_existing_dataset() -> None:
+    delete_dir(TESTOUTPUT_DIR / "simple_wk_dataset")
+    ds1 = Dataset.create(TESTOUTPUT_DIR / "simple_wk_dataset", scale=(1, 1, 1))
+    ds1.add_layer(
+        "color", LayerCategories.COLOR_TYPE, dtype_per_layer="float", num_channels=1
+    )
+    
+    ds2 = Dataset(TESTOUTPUT_DIR / "simple_wk_dataset")
+    ds2.add_layer(
+        'segmentation', LayerCategories.SEGMENTATION_TYPE, 'uint8', largest_segment_id=100000
+    )
+    
+    assert((TESTOUTPUT_DIR / "simple_wk_dataset" / "segmentation").is_dir())
 
 
 def test_view_read_with_open() -> None: