Identify on arcpy.layers.MapImageLayer not working because of nested dictionary in geometry

[tbock]

Describe the bug
Identify on MapImagerLayer return response saying geometry was not provided when it was. You appear to have fallen victim to the well know issue of python requests and nested dictionaries.

To Reproduce
Steps to reproduce the behavior:

l = arcgis.layers.MapImageLayer('https://coast.noaa.gov/arcgis/rest/services/FloodExposureMapper/CFEM_CoastalFloodHazardComposite/MapServer', gis=gis)
f = {"geometry": Point({
        "x" : -13575713.5781, 
        "y" : 4384575.0913000032, 
    "spatialReference": {"wkid": 102100, "latestWkid": 3857}})}

l.identify(geometry=f['geometry'], 
           geometry_type='Point',     
           map_extent={       
               'xmin': f['geometry']['x'],
               'xmax': f['geometry']['x'],
               'ymin': f['geometry']['y'],
               'ymax': f['geometry']['y'],
                'spatialReference': site['geometry']['spatialReference']
            }, 
           image_display='600,550,96',
           tolerance=0,
           return_geometry=False,
           layers='visible')

error:

{'error': {'code': 400,
  'message': "Missing 'geometry' for identify operation.",
  'details': []}}

Screenshots
You can clearly see the mangled data that requests is sending to the server.

Expected behavior
Expect it to return the output of the identify operation from the service.

Platform (please complete the following information):

• OS: Windows and ArcGIS Online
• Browser n/a and Chrome
• Python API Version 2.4.0

Additional context
I should not have to wrap the geometry in json.dumps to get it to work.

l.identify(geometry=json.dumps(f['geometry']), 
           geometry_type='Point',     
           map_extent=json.dumps({       
               'xmin': f['geometry']['x'],
               'xmax': f['geometry']['x'],
               'ymin': f['geometry']['y'],
               'ymax': f['geometry']['y'],
                'spatialReference': site['geometry']['spatialReference']
            }), 
           image_display='600,550,96',
           tolerance=0,
           return_geometry=False,
           layers='visible')

[nanaeaubry]

@tbock Thanks for reporting this, we will take a look!