In a preprocessing step, I want to flush a few states from a TωA.  The one method that seems to be available is defrag_states, the description of which having the expected semantics.

However, after defrag, it sometimes happens that the dst of some transitions points to delete states (-1U).  The code of defrag_states in graph.hh removes the outgoing transitions of deleted states but does not seem to flush the transitions pointing to deleted states—is this the expected behavior?

Cheers,

M.