.to_rioxarray method for async_geotiff._array.Array

[raybell-md]

I ended up writing a util in #92 (comment)

To convert an async_geotiff._array.Array to an xarray.DataArray.

Open to adding the util here. No preference if this util (along with many other utils) should stay with the user as not to bloat this project.

[weiji14]

We've discussed internally whether to add an xarray BackendEntrypoint (answer is: not sure 😆). We would need to be careful to add xarray as an optional dependency so that the core package remains unbloated.

And just in case you're interested, I wrote a similar to_xarray() method at https://github.com/developmentseed/pixelverse/pull/24/changes#diff-8feec44dae0c09e9044a272dadfb9775d686299a5822d4ca7c62bc8531f7a2dd, though this was using async-tiff instead of async-geotiff (haven't got around to rewrite it yet), and the coordinates was using rasterix (for some lazy indexing).

[kylebarron]

I think without adding a whole BackendEntrypoint, having a to_xarray method on Array could be in scope (xarray would be a conditional dependency)

[kylebarron]

rasterix looks really nice. Thanks @dcherian!

So it should be one line of code to create an xarray indexer from an Array:

from rasterix import RasterIndex

index = RasterIndex.from_transform(
	array.transform,
	width=array.width,
	height=array.height,
	crs=array.crs,
)

Then one line of code to assign that to your array

# Assign to data
ds_manual = xr.Dataset(
    {"data": (("y", "x"), np.random.rand(100, 100))},
    coords=xr.Coordinates.from_xindex(index),
)
ds_manual

[dcherian]

rasterix looks really nice.

ahem... it looks.... rasterized. 🥁

[kylebarron]

Put up one thought here: #95