It would be nice to match behaviour of NumPy and have e.g. ``` a = blosc2.linspace(0, 10, num=100, shape=(10,10)) print(a[3, None].shape) # has shape (1,10) ``` so that new dummy axes may be introduced using ```None```.