Parser expects graph wrapped

[clausmichele]

@GeraldIr I noticed an issue with the changes you did here:
2023.10.5...2023.10.6

After the changes, with release 2023.10.6, when a process graph contains fit_curve or predict_curve, it expects to be wrapped with a "process_graph" key, which is not the case when creating process graphs with the Python Client.

I found this bug while trying to use fit_curve locally. Here is where the graph is being parsed: https://github.com/Open-EO/openeo-python-client/blob/522edb7ffad4a14f27536a060eac551010d8312a/openeo/local/connection.py#L275

This sample code works with 2023.10.5 and no more with 2023.10.6:

import openeo
from openeo.rest.datacube import PGNode, THIS
from openeo.processes import sin, cos
import math

backend = "openeo.cloud"
conn = openeo.connect(backend)
conn = conn.authenticate_oidc()

def fit_process(bbox):
    spatial_extent  = {'west':bbox[0],'east':bbox[2],'south':bbox[1],'north':bbox[3]}
    temporal_extent = ["2016-09-01", "2018-09-01"] 
    collection      = 'boa_sentinel_2'
    bands           = ["B02", "B03", "B04", "B05", "B08"] 

    l2a_bands = conn.load_collection(collection,spatial_extent=spatial_extent,bands=bands,temporal_extent=temporal_extent)

    l2a_bands_clipped = l2a_bands.apply(lambda x: x.clip(0,5000))

    def fitFunction(x, parameters):
        t = 2 * math.pi / 31557600 * x
        return parameters[0] + parameters[1] * cos(t) + parameters[2] * sin(t)

    curve_fitting = l2a_bands_clipped.fit_curve(
        parameters=[1,1,1],
        dimension="t",
        function=fitFunction
    )
    fitted_parameters = curve_fitting.save_result(format="NetCDF", options={"tile_grid":"time-series"})
    return fitted_parameters

bbox = [11.65, 46.3, 11.7, 46.35]
fitted_parameters = fit_process(bbox)

from openeo_pg_parser_networkx import OpenEOProcessGraph
OpenEOProcessGraph(fitted_parameters.flat_graph())