collapseReplicates bugged with sparse matrices

[arturspeiser]

Using the example:

ddse = makeExampleDESeqDataSet(m=12)
ddse.X =  scipy.sparse.csr_matrix(ddse.X)
# make data with two technical replicates for three samples
ddse.obs["sample"] = inmoose.utils.Factor(np.repeat(np.arange(1,10), [2,1,1,2,1,1,2,1,1]))
ddse.obs["run"] = [f"run{i}" for i in range(12)]
ddsColl = collapseReplicates(ddse, ddse.obs["sample"], ddse.obs["run"])

returns

ValueError: Data matrix has wrong shape (9, 1, 1000), need to be (9, 1000).