bug: indexing length 1 table with bool array

[BastiaanBergman]

When indexing a length 1 table with a bool array, the index is treated a s a single integer number. This is because int(np.array([False])) is treated as 0 and will not raise an error inside the Tabel.getitem method.

Example code:

# Correct output
tbl = Tabel({'a':[1,1.1],'b':[2,2.2]})
tbl[tbl['a']>2, :]
>>>     a   | b
>>>    -----+-----
>>>    0 rows ['<f8', '<f8']

# Incorrect output
tbl = Tabel({'a':[1],'b':[2]})
tbl[tbl['a']>2, :]
>>>    (1, 2)

# Workaround:
tbl = Tabel({'a':[1],'b':[2]})
tbl[tuple(tbl['a']>2), :]
>>>      a   | b
>>>     -----+-----
>>>     0 rows ['<i8', '<i8']

[BastiaanBergman]

Actually, the workaround is not good. Some numpy versions don't allow bool indexing with tubples.

[BastiaanBergman]

To prevent users from doing this we need to implement python type checking with type hints.