remove ugly adding hack since we handle it differently in boost histo…

…gram now

src/dask_histogram/core.py

@@ -603,9 +603,7 @@ def counts(self, flow: bool = False) -> NDArray[Any]:
     def __array__(self) -> NDArray[Any]:
         return self.compute().__array__()
 
-    def __iadd__(self, other: Any) -> AggHistogram:
-        if isinstance(other, AggHistogram):
-            return _iadd_hist(self, other)
+    def __iadd__(self, other: Any) -> AggHistogram:        
         return _iadd(self, other)
 
     def __add__(self, other: Any) -> AggHistogram:
@@ -1058,91 +1056,6 @@ def __call__(self, a: AggHistogram, b: AggHistogram) -> AggHistogram:
         return AggHistogram(g, name, histref=ref)
 
 
-class UnorderedTreeReductionBinaryOp:
-    def __init__(
-        self,
-        func: Callable[[Any], Any],
-        name: str | None = None,
-    ) -> None:
-        self.func = func
-        self.__name__ = func.__name__ if name is None else name
-
-    def __call__(self, a: AggHistogram, b: AggHistogram) -> AggHistogram:
-        token = tokenize(a, b)
-        name = f"{self.__name__}-hist-{token}"
-        name_comb = f"{self.__name__}-combine-{token}"
-        deps = []
-        if is_dask_collection(a):
-            deps.append(a)
-        if is_dask_collection(b):
-            deps.append(b)
-
-        layer_a = (
-            [
-                value
-                for value in a.dask.layers.values()
-                if isinstance(value, MockableMultiSourceTreeReduction)
-            ]
-            if is_dask_collection(a)
-            else a
-        )
-        layer_b = (
-            [
-                value
-                for value in b.dask.layers.values()
-                if isinstance(value, MockableMultiSourceTreeReduction)
-            ]
-            if is_dask_collection(b)
-            else b
-        )
-        layer_a = layer_a[0] if len(layer_a) == 1 else None
-        layer_b = layer_b[0] if len(layer_b) == 1 else None
-
-        layer_a_names = layer_a.names_inputs if layer_a else tuple()
-        layer_a_parts = layer_a.npartitions_inputs if layer_a else tuple()
-        layer_b_names = layer_b.names_inputs if layer_b else tuple()
-        layer_b_parts = layer_b.npartitions_inputs if layer_b else tuple()
-
-        a_b_reduction = MockableMultiSourceTreeReduction(
-            name=name,
-            names_inputs=(layer_a_names + layer_b_names),
-            npartitions_inputs=(layer_a_parts + layer_b_parts),
-            concat_func=self.func,
-            tree_node_func=layer_a.tree_node_func,
-            finalize_func=layer_a.finalize_func,
-            split_every=layer_a.split_every,
-            tree_node_name=name_comb,
-        )
-        layers = {name: a_b_reduction}
-        name_dep = set(a_b_reduction.names_inputs)
-        deps = {name: name_dep}
-        layers_a = a.dask.layers.copy()
-        if layer_a:
-            layers_a.pop(a.name)
-        layers_b = b.dask.layers.copy()
-        if layer_b:
-            layers_b.pop(b.name)
-        layers.update(layers_a)
-        layers.update(layers_b)
-
-        a_deps = a.dask.dependencies.copy()
-        a_deps.pop(a.name)
-        b_deps = b.dask.dependencies.copy()
-        b_deps.pop(b.name)
-
-        deps.update(a_deps)
-        deps.update(b_deps)
-
-        g = HighLevelGraph(layers, deps)
-        try:
-            ref = a.histref
-        except AttributeError:
-            ref = b.histref
-
-        return AggHistogram(g, name, histref=ref)
-
-
-_iadd_hist = UnorderedTreeReductionBinaryOp(hist_safe_sum, "add")
 _iadd = BinaryOpAgg(operator.iadd, name="add")
 _isub = BinaryOpAgg(operator.isub, name="sub")
 _imul = BinaryOpAgg(operator.imul, name="mul")