FiniteStateMachine: colors lists, not tuples

[cheuberg]

Between the time when #15963 was merged in develop and #15963 was actually closed, I found and fixed a bug and pushed it to #15963, see #15963 comment:4. However, due to the race condition outlined above, this fix has never been merged.

Thus, this is a new ticket with the same fix.

For convenience, I repeat #15963 comment:4:

The following behaviour is undesired (docstrings says that colors are tuples of the colors of the constituent states, but this is not the case and leads to problems in Automaton.determinisation):

sage: A = Automaton([[0, 0, 0]], initial_states=[0])
sage: B = A.product_FiniteStateMachine(A,
....:                                  lambda t1, t2: (0, None))
sage: B.states()[0].color
[None, None]
sage: B.determinisation()
Traceback (most recent call last):
...
TypeError: unhashable type: 'list'
sage: A = Automaton([[0, 0, 0]], initial_states=[0])
sage: B = A.composition(A, algorithm='explorative')
sage: B.states()[0].color
[None, None]
sage: B.determinisation()
Traceback (most recent call last):
...
TypeError: unhashable type: 'list'

I pushed commits correcting this and adding two more docstrings to emphasize that colors have to be hashable in order to be able to use Automaton.determinisation.

CC: @dkrenn @sagetrac-skropf

Component: combinatorics

Author: Clemens Heuberger

Branch/Commit: d51be51

Reviewer: Daniel Krenn

Issue created by migration from https://trac.sagemath.org/ticket/16128

[dkrenn]

Reviewer: Daniel Krenn

[dkrenn]

comment:2

I've checked already the changes between merging and closing of #15963. Everything looks fine.

[vbraun]

Changed branch from u/cheuberg/fsm/state_color to d51be51