⏱️ Find quicker way to detect a change in a roadway network

[e-lo]

Right now wrangler stores a roadway network-hash every time it changes its value in order to detect if other operations need to be rerun or can use stored values (i.e. selections, model network changes, etc).

The hash is a hash of the combined hashes of links and node dataframes which is a hash of the underlying numpy array. This is taking about a second each time it is created for just the ST Paul network, which adds up. This is already only done lazily when it is actually needed so we would need to speed up the actual hash creation.

@property
def network_hash(self) -> str:
    """Hash of the links and nodes dataframes."""
    _value = str.encode(self.links_df.df_hash() + "-" + self.nodes_df.df_hash())

    _hash = hashlib.sha256(_value).hexdigest()
    return _hash

@pd.api.extensions.register_dataframe_accessor("df_hash")
class dfHash:
"""Creates a dataframe hash that is compatable with geopandas and various metadata.

Definitely not the fastest, but she seems to work where others have failed.
"""

def __init__(self, pandas_obj):
    """Initialization function for the dataframe hash."""
    self._obj = pandas_obj

def __call__(self):
    """Function to hash the dataframe."""
    _value = str(self._obj.values).encode()
    hash = hashlib.sha1(_value).hexdigest()
    return hash