load_stac enhancements

[clausmichele]

This PR will include many enhancements for the load_stac process:

• Switch from stackstac to odc.stac, see the reasons here: load_stac: possible improvements #235
• Increase support for other data types. With this PR, it's possible to load Zarr data, see this example: https://gist.github.com/clausmichele/ba9fdbf5add71504b7553f7f7b6d1be0
• Maintain the source data type instead of converting everything to float64, reducing the memory and computational resources required.
• Get the dimension names from the STAC datacube extension, if available (still not sure about this, need to discuss)

If you have any other suggestion of possible improvements, please comment