grid metrics when first creating the oceandataset

Description

Should we allow the grid object to be created incorporating the grid metrics?

So far, metrics have not been used to create and open the od object in any of the dataset. You can see this in oceanpy.ospy_utils (lines 15-33):

def _create_grid(dataset, coords, periodic, face_connections):
    """
    Create xgcm grid by adding comodo attributes to the
    dimensions of the dataset.

    Parameters
    ----------
    dataset: xarray.Dataset
    coords: dict
        E.g., {'Y': {Y: None, Yp1: 0.5}}
    periodic: list
        List of periodic axes.
    face_connections: dict
        dictionary specifying grid topology
    Returns
    -------
    grid: xgcm.Grid
    """

and in lines 60-66:

        grid = xgcm.Grid(dataset, periodic=periodic, face_connections=face_connections)

Incorporating grid metrics is advantageous and very simple. From xgcm's documentation:

metrics = {
    ('X',): ['dxC', 'dxG'], # X distances
    ('Y',): ['dyC', 'dyG'], # Y distances
    ('Z',): ['drW', 'drS', 'drC'], # Z distances
    ('X', 'Y'): ['rA', 'rAz', 'rAs', 'rAw'] # Areas
}
grid = Grid(ds, metrics=metrics)

Allowing the metrics to be passed into Grid allow one-line calculation of derivatives, integrals and such. This is:

With Grid metrics:
```
U_int1= grid.integrate(ds.UVEL, 'Z')
```
Without Grid metrics:
```
U_int2 = (ds.UVEL * ds.drW).sum('Z')
```
In reality U_int1 == U_int2.

hainegroup / oceanspy

grid metrics when first creating the oceandataset #298

Description