bc.st.clr_normalize() fails

Running standard_workflow notebooks fails when running with CITEseq data

```
ValueError Traceback (most recent call last)
Cell In[38], line 30
28 #perform normalization (this normalization is specific to CITEseq data)
29 if adt_norm == 'CLR':
---> 30 bc.st.clr_normalize(adata_prot, os.path.join(results_folder_citeseq, 'citeseq'))
32 #perform batch correction if desired otherwise just perform clustering
33 if (batch_to_correct != 'None'):
34 #save a copy of uncorrected in case we need it for something later

File .../lib/python3.9/site-packages/besca/st/_wrapper_funcs.py:290, in clr_normalize(adata, results_folder)
285 start = time()
287 # normalize per cell
288 # this also already applies log! Is not taken seperately
289 # already normalize BEFORE saving "raw" - as recommended in the scanpy tutorial
--> 290 normalize_geometric(adata)
291 print("clr normalization applied to adata")
293 # keep raw copy

File .../lib/python3.9/site-packages/besca/pp/_normalization.py:188, in normalize_geometric(adata)
185 X = np.array(X)
187 # replacement of zero values with very small numbers without changing the overall sums
--> 188 X = multiplicative_replacement(X)
190 # centre log ratio transformation
191 X = clr(X)

File .../lib/python3.9/site-packages/besca/pp/_normalization.py:132, in multiplicative_replacement(mat, delta)
87 def multiplicative_replacement(mat, delta=None):
88 r"""Replace all zeros with small non-zero values
89 It uses the multiplicative replacement strategy [1]_ ,
90 replacing zeros with a small positive :math:`\delta`
(...)
129 [0.0625, 0.4375, 0.4375, 0.0625]])
130 """
--> 132 mat = closure(mat)
133 z_mat = mat == 0
135 num_feats = mat.shape[-1]

File .../lib/python3.9/site-packages/besca/pp/_normalization.py:37, in closure(mat)
6 """
7 Performs closure to ensure that all elements add up to 1.
8
(...)
34 [0.4, 0.4, 0.2]])
35 """
36 mat = np.atleast_2d(mat)
---> 37 if np.any(mat < 0):
38 raise ValueError("Cannot have negative proportions")
39 if mat.ndim > 2:

File .../lib/python3.9/site-packages/scipy/sparse/_base.py:396, in _spbase.__bool__(self)
394 return self.nnz != 0
395 else:
--> 396 raise ValueError("The truth value of an array with more than one "
397 "element is ambiguous. Use a.any() or a.all().")

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all().
```

Start with besca/st/_wrapper_funcs.py at clr_normalize and follow normalize_geometric, multiplicative_replacement, and closure in besca/pp/_normalization.py. Reproduce the failure by running the standard_workflow notebooks with CITEseq data, then verify that CLR normalization completes without the reported ValueError.

Written by the indexing model from the issue text.