Cannot flatten lists of tuples containing lists

[maffoo]

When flattening python data into labrad format, pylabrad tries to infer the correct python type as it goes. This can become complicated when flattening lists of more complicated structures, because lists must be homogeneous so that each element has the same type, but the final type may need to be refined as we visit more elements in the list. We ran into the following edge case, where we want to flatten a list of clusters of two lists of strings:

>>> data1 = [([],), (['a'],), ([],)]
>>> print(labrad.types.flatten(data1).tag)
*(*s)

>>> data2 = [([], []), ([], ['a']), (['a'], [])]
>>> print(labrad.types.flatten(data2).tag)
FlatteningError: (12345) Could not flatten 'a' to _. [payload=none]

>>> print(labrad.types.flatten(data2, '*(*s*s)').tag
*(*s*s)

The first example with clusters containing one list works fine; the flattening code infers that the list contains strings after it sees the second element. But the next example with clusters containing two lists does not work as expected. If we provide an explicit type as in the last example, everything flattens just fine, but it should work even if we don't provide an explicit type to flatten to.