Build: fix build in mypyc >=1.14

[tlambert03]

Not sure what happened yet... but something changed in mypyc that causes compile failures, and persistent test failures that look like this:

    def test_group_blocked_exclude() -> None:
        """Test that we can exempt certain signals from being blocked."""
        group = MyGroup()

        mock1 = Mock()
        mock2 = Mock()

        group.sig1.connect(mock1)
        group.sig2.connect(mock2)

>       with group.all.blocked(exclude=("sig2",)):

tests/test_group.py:172:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
src/psygnal/_signal.py:1502: in __enter__
    self._signal.block(exclude=self._exclude)
src/psygnal/_group.py:162: in block
    if (exclude and sig in exclude) or name in exclude:
src/psygnal/_signal.py:1130: in __contains__
    return self._slot_index(slot) >= 0
src/psygnal/_signal.py:1094: in _slot_index
    normed = weak_callback(slot, on_ref_error="ignore")
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

>   raise TypeError(f"unsupported type {type(cb)}")  # pragma: no cover
E   TypeError: unsupported type <class 'tuple'>

weird thing is it seems like it's flipping around the arguments to sig in exclude, such that it runs at runtime as if it were exclude in sig?

i'm going to pin it for now, but opening the issue so I don't forget