Cell instances should contain big brain coordinates as well

             x_um       y_um   area_um2
cell_id
1       -3.622802 -12.077647  25.524589
...           ...        ...        ...

siibra-python requires x,y,z values for digesting point clouds such as cell distiributions. It makes sense for x_um and y_um to stay but would be great to have coordinates in bigbrain space. It can have x_bigbrain_mm type of column names which I can map to x,y,z. Needs a few changes on siibra-python but not so difficult.

So given a tsv

             x_um       y_um   area_um2  x_bigbrain_mm y_bigbrain_mm  z_bigbrain_mm
cell_id
1       -3.622802 -12.077647  25.524589   1  4  5
...           ...        ...        ...   23  2  55

The reader could look like:

pd.read_csv('./test.csv', index_col=0, skiprows=1, names=['cell id', 'x_patch_um', 'y_patch_um', 'area_um2', 'x', 'y', 'z'])

which could be in configuration json as:

"decoder": {
    "@type": "siibra/decoder/csv",
    "delimiter": "\\t",
    "engine": "python",
    "header": 0,
    "index_col": 0,
    "skiprows": 1
}