Extract D-inf flow accumulation into dedicated module and update hydrology API

- Move D-inf flow accumulation from flow_accumulation.py into
  flow_accumulation_dinf.py with its own test file
- Update basin, watershed, stream_order, and stream_order_mfd to use
  D8 flow direction consistently
- Add flow_accumulation_dinf to public API and docs
- Simplify flow_accumulation.py to D8-only logic

Author: brendancol

docs/source/reference/hydrology.rst

@@ -32,6 +32,13 @@ Flow Accumulation
 
     xrspatial.flow_accumulation.flow_accumulation
 
+Flow Accumulation (D-infinity)
+===============================
+.. autosummary::
+    :toctree: _autosummary
+
+    xrspatial.flow_accumulation_dinf.flow_accumulation_dinf
+
 Flow Accumulation (MFD)
 =======================
 .. autosummary::

xrspatial/__init__.py

@@ -39,6 +39,7 @@
 from xrspatial.flood import inundation  # noqa
 from xrspatial.flood import travel_time  # noqa
 from xrspatial.flow_accumulation import flow_accumulation  # noqa
+from xrspatial.flow_accumulation_dinf import flow_accumulation_dinf  # noqa
 from xrspatial.flow_accumulation_mfd import flow_accumulation_mfd  # noqa
 from xrspatial.flow_direction import flow_direction  # noqa
 from xrspatial.flow_direction_dinf import flow_direction_dinf  # noqa

xrspatial/basin.py

@@ -364,7 +364,10 @@ def basin(flow_dir: xr.DataArray,
         fd = data.astype(np.float64)
         h, w = fd.shape
         labels = _basins_init_labels(fd, h, w, h, w, 0, 0)
-        out = _watershed_cpu(fd, labels, h, w)
+        # Build state array: 0=nodata(NaN), 1=unresolved(-1), 3=resolved
+        state = np.where(np.isnan(labels), 0,
+                         np.where(labels == -1.0, 1, 3)).astype(np.int8)
+        out = _watershed_cpu(fd, labels, state, h, w)
 
     elif has_cuda_and_cupy() and is_cupy_array(data):
         out = _basins_cupy(data)