Faster subsetting for cohorts (#397)

* Faster subsetting for cohorts

Closes #396

* tpying

flox/core.py

@@ -1494,8 +1494,9 @@ def _normalize_indexes(array: DaskArray, flatblocks, blkshape) -> tuple:
 def subset_to_blocks(
     array: DaskArray,
     flatblocks: Sequence[int],
-    blkshape: tuple[int] | None = None,
+    blkshape: tuple[int, ...] | None = None,
     reindexer=identity,
+    chunks_as_array: tuple[np.ndarray, ...] | None = None,
 ) -> DaskArray:
     """
     Advanced indexing of .blocks such that we always get a regular array back.
@@ -1518,6 +1519,9 @@ def subset_to_blocks(
     if blkshape is None:
         blkshape = array.blocks.shape
 
+    if chunks_as_array is None:
+        chunks_as_array = tuple(np.array(c) for c in array.chunks)
+
     index = _normalize_indexes(array, flatblocks, blkshape)
 
     if all(not isinstance(i, np.ndarray) and i == slice(None) for i in index):
@@ -1531,7 +1535,7 @@ def subset_to_blocks(
     new_keys = array._key_array[index]
 
     squeezed = tuple(np.squeeze(i) if isinstance(i, np.ndarray) else i for i in index)
-    chunks = tuple(tuple(np.array(c)[i].tolist()) for c, i in zip(array.chunks, squeezed))
+    chunks = tuple(tuple(c[i].tolist()) for c, i in zip(chunks_as_array, squeezed))
 
     keys = itertools.product(*(range(len(c)) for c in chunks))
     layer: Graph = {(name,) + key: (reindexer, tuple(new_keys[key].tolist())) for key in keys}
@@ -1726,14 +1730,15 @@ def dask_groupby_agg(
 
             reduced_ = []
             groups_ = []
+            chunks_as_array = tuple(np.array(c) for c in array.chunks)
             for blks, cohort in chunks_cohorts.items():
                 cohort_index = pd.Index(cohort)
                 reindexer = (
                     partial(reindex_intermediates, agg=agg, unique_groups=cohort_index)
                     if do_simple_combine
                     else identity
                 )
-                reindexed = subset_to_blocks(intermediate, blks, block_shape, reindexer)
+                reindexed = subset_to_blocks(intermediate, blks, block_shape, reindexer, chunks_as_array)
                 # now that we have reindexed, we can set reindex=True explicitlly
                 reduced_.append(
                     tree_reduce(