Tuple.reduce with ellipses raises on the last invalid index

[asmeurer]

For example:

>>> Tuple(..., 0, 0).reduce((0, 0))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/aaronmeurer/Documents/ndindex/ndindex/tuple.py", line 367, in reduce
    s = s.reduce(shape, axis=axis, negative_int=negative_int)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/aaronmeurer/Documents/ndindex/ndindex/integer.py", line 100, in reduce
    self._raise_indexerror(shape, axis)
  File "/Users/aaronmeurer/Documents/ndindex/ndindex/integer.py", line 63, in _raise_indexerror
    raise IndexError(f"index {self.raw} is out of bounds for axis {axis} with size {size}")
IndexError: index 0 is out of bounds for axis 1 with size 0

vs.

>>> np.empty((0, 0))[..., 0, 0]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
IndexError: index 0 is out of bounds for axis 0 with size 0

This means we have to set same_exception=False all over the tests.

It's because of the reverse loop

ndindex/ndindex/tuple.py

Line 337 in bdb7edb

for i, s in enumerate(reversed(args[:ellipsis_i]), start=1):

[asmeurer]

Maybe we could raise them all together with an exception group. Not sure if it's a good idea, though.