CallContext bug leads to context values bleeding

[dualc]

I found some bugs about CallContext when I changed this jupyter to suppor multi-user/tenancy ，

@classmethod
def _get_map(cls) -> Dict[str, Any]:
"""Get the map of names to their values from the _NAME_VALUE_MAP context var.

    If the map does not exist in the current context, an empty map is created and returned.
    """
    ctx: Context = copy_context()
    if CallContext._name_value_map not in ctx:
        CallContext._name_value_map.set({})
    return CallContext._name_value_map.get()

copy_context is a shallow copy，_name_value_map is a dict，so when multi user visite jupyter，if we record some user infomation in context , it will casue context values bleeding

[dualc]

I found a solution to fix it. define a reset method in CallContext

    @classmethod
    def reset(cls):
        CallContext._name_value_map.set({})

when handler prepare，call the reset to set a dict to context

    async def prepare(self, *, _redirect_to_login=True) -> Awaitable[None] | None:  # type:ignore[override]
        """Prepare a response."""
        # Set the current Jupyter Handler context variable.
        CallContext.reset()
        CallContext.set(CallContext.JUPYTER_HANDLER, self)