Are subtypes of Cartan types implemented correctly?
Consider the following code
CartanType("B3").subtype([2, 3]).index_set()
CartanType("B3").subtype([1, 3]).index_set()
It gives
(2, 3)
(1, 2)
in Sage 8.4. Is it wrong to expect the index set of a subtype to be the subset used in defining the subtype? From my point of view this inconsistency leads to all kinds of unexpected behaviour in the relabellings for subtypes, but maybe it is intentional?