tiny doc cleanups

might as well polish these while I’m here

python-spec/src/somacore/query/_fast_csr.py

@@ -61,10 +61,11 @@ def to_scipy(self) -> sparse.csr_matrix:
             sparse.csr_matrix((data, indices, indptr), shape=shape)
 
         This ugliness is to bypass the O(N) scan that
-        :meth:`scipy.sparse._cs_matrix.__init__`
+        :meth:`sparse._cs_matrix.__init__`
         does when a new compressed matrix is created.
 
-        See https://github.com/scipy/scipy/issues/11496 for details on the bug.
+        See `SciPy bug 11496 <https://github.com/scipy/scipy/issues/11496>`
+        for details.
         """
         matrix = sparse.csr_matrix.__new__(sparse.csr_matrix)
         matrix.data = self.data
@@ -114,6 +115,7 @@ def append(
     ) -> None:
         """
         At accumulation time, do several things:
+        
         * re-index to positional indices, and if possible, cast to smaller dtype
           to minimize memory footprint (at cost of some amount of time)
         * accumulate column counts by row, i.e., build the basis of the indptr