"Cannot open data source" error after dropna

I use the following lines to open shapefile as GeoDataFrame, drop NaN values and use the final GeoDataFrame without NaN values in pygeoda. But after `dropna` pygeoda throws this error: `ValueError: pygeoda can't open current data source. Please use either a file path of an ESRI shapefile or a GeoPandas instance.`, However, both `gdf1` and `gdf2` are GeoDataFrame with many rows.

```
gdf1 = gpd.read_file("/home/user/test.shp")
print(type(gdf1))
# OUT
data1 = pygeoda.open(gdf1) # No Error here

gdf2 = gdf1.dropna()
print(type(gdf2))
# OUT
data2 = pygeoda.open(gdf2) # ERROR
```

geopandas v0.9.0 and 0.10.0
pygeoda v0.0.8-1

Reproduce the example using pygeoda.open, first with the original GeoDataFrame and then with the result of dropna, using the reported GeoPandas and pygeoda versions. Compare how each GeoDataFrame is accepted and identify what prevents the filtered instance from opening. Done means the dropna result can be passed to pygeoda.open without the reported ValueError, with coverage for this case.

Written by the indexing model from the issue text.