Can't mix variables from obs and var in pl.embedding

[grst]

Describe the bug
When trying to plot variables from obs and var in the same pl.embedding, it fails.

To Reproduce

import muon as mu
import scanpy as sc

mdata = mu.MuData({"gex": sc.datasets.pbmc3k_processed()})

# works
mu.pl.embedding(mdata, "gex:umap", color=["CD8A"])
# works
mu.pl.embedding(mdata, "gex:umap", color=["gex:louvain"])
# Key Error
mu.pl.embedding(mdata, "gex:umap", color=["gex:louvain", "CD8A"])

KeyError: 'Could not find key louvain in .var_names or .obs.columns.'

Stacktrace

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
Cell In[31], line 1
----> 1 mu.pl.embedding(mdata, "gex:umap", color=["gex:louvain", "CD8A"])

File ~/anaconda3/envs/scirpy_dev/lib/python3.9/site-packages/muon/_core/plot.py:255, in embedding(data, basis, color, use_raw, layer, **kwargs)
    252     color = [mod_key_modifier[k] for k in keys]
    254 ad = AnnData(obs=obs, obsm=adata.obsm, obsp=adata.obsp, uns=adata.uns)
--> 255 return sc.pl.embedding(ad, basis=basis_mod, color=color, **kwargs)

File ~/anaconda3/envs/scirpy_dev/lib/python3.9/site-packages/scanpy/plotting/_tools/scatterplots.py:244, in embedding(adata, basis, color, gene_symbols, use_raw, sort_order, edges, edges_width, edges_color, neighbors_key, arrows, arrows_kwds, groups, components, layer, projection, scale_factor, color_map, cmap, palette, na_color, na_in_legend, size, frameon, legend_fontsize, legend_fontweight, legend_loc, legend_fontoutline, vmax, vmin, vcenter, norm, add_outline, outline_width, outline_color, ncols, hspace, wspace, title, show, save, ax, return_fig, **kwargs)
    235 # use itertools.product to make a plot for each color and for each component
    236 # For example if color=[gene1, gene2] and components=['1,2, '2,3'].
    237 # The plots are: [
    238 #     color=gene1, components=[1,2], color=gene1, components=[2,3],
    239 #     color=gene2, components = [1, 2], color=gene2, components=[2,3],
    240 # ]
    241 for count, (value_to_plot, component_idx) in enumerate(
    242     itertools.product(color, idx_components)
    243 ):
--> 244     color_source_vector = _get_color_source_vector(
    245         adata,
    246         value_to_plot,
    247         layer=layer,
    248         use_raw=use_raw,
    249         gene_symbols=gene_symbols,
    250         groups=groups,
    251     )
    252     color_vector, categorical = _color_vector(
    253         adata,
    254         value_to_plot,
   (...)
    257         na_color=na_color,
    258     )
    260     # Order points

File ~/anaconda3/envs/scirpy_dev/lib/python3.9/site-packages/scanpy/plotting/_tools/scatterplots.py:1225, in _get_color_source_vector(adata, value_to_plot, use_raw, gene_symbols, layer, groups)
   1223     values = adata.raw.obs_vector(value_to_plot)
   1224 else:
-> 1225     values = adata.obs_vector(value_to_plot, layer=layer)
   1226 if groups and is_categorical_dtype(values):
   1227     values = values.replace(values.categories.difference(groups), np.nan)

File ~/anaconda3/envs/scirpy_dev/lib/python3.9/site-packages/anndata/_core/anndata.py:1375, in AnnData.obs_vector(self, k, layer)
   1369         warnings.warn(
   1370             "In a future version of AnnData, access to `.X` by passing"
   1371             " `layer='X'` will be removed. Instead pass `layer=None`.",
   1372             FutureWarning,
   1373         )
   1374         layer = None
-> 1375 return get_vector(self, k, "obs", "var", layer=layer)

File ~/anaconda3/envs/scirpy_dev/lib/python3.9/site-packages/anndata/_core/index.py:191, in get_vector(adata, k, coldim, idxdim, layer)
    187     raise ValueError(
    188         f"Key {k} could be found in both .{idxdim}_names and .{coldim}.columns"
    189     )
    190 elif (in_col + in_idx) == 0:
--> 191     raise KeyError(
    192         f"Could not find key {k} in .{idxdim}_names or .{coldim}.columns."
    193     )
    194 elif in_col:
    195     return getattr(adata, coldim)[k].values

KeyError: 'Could not find key louvain in .var_names or .obs.columns.'

Expected behaviour
Being able to mix features from obs and var in the same embedding plot

System
muon 0.1.3
scanpy 1.8.2
mudata 0.2.2

Additional context
This still worked as expected in muon 0.1.2