Cannot handle obsm data with >2 dimensions

[human9]

Issue summary

AnnData only requires entries in obsm have "a first dimension of size n_obs", but AnndataR seems to additionally require that the number of dimensions is no larger than 2.

Example: reading h5ad

For example, I have an h5ad containing the following 3D obsm entry which I can load up in Python:

> import anndata
> adata = anndata.read_h5ad("spatial_data.h5ad")
> adata.obsm['cell_border'].shape

(130597, 32, 2)

I try to load it in anndataR:

> library(anndataR)
> adata <- read_h5ad("spatial_data.h5ad")

Error in private$.validate_aligned_array(collection[[mtx_name]], collection_name,  : 
  dim(obsm[['cell_border']])[1] should have shape: 130597, found: 2.

Example: saving h5ad

Also, I can create an object with some n-dimensional obsm in AnndataR manually, but I can't save it back to a file because it requires the entry to be a matrix:

adata <- AnnData(
  X = matrix(rnorm(100), nrow = 10),
  obs = data.frame(
    cell_type = factor(rep(c("A", "B"), each = 5))
  ),
  var = data.frame(
    gene_name = paste0("gene_", 1:10)
  ),
  obsm = list(borders = array(rep(0, 10*2*10), c(10, 2, 10))),
)
adata$to_HDF5AnnData("myanndata.h5ad")

Warning message:
  In value[[3L]](cond) :
  Could not write element 'obsm/borders' of type 'array':
  argument is not a matrix