Class Ruleset

Ruleset class, used to compute NFA and then DFA for parsing based on
a list of rules.

compute DFA for ruleset by computing the E-closure of the NFA

method to compute prefixes and First sets for nonterminals

computing the Follow set for the ruleset
the good news: I think it's correct.
the bad news: It's slower than it needs to be for epsilon cases.

compute an SLR NFA for the ruleset with states for each SLR "item"
and transitions, eg:
    X > .AB
  on A maps to X > A.B
  on epsilon maps to A > .ZC
                 and A > .WK
an item is a pair (rulenumber, bodyposition)
where body position 0 is interpreted to point before the
beginning of the body.

SLR = "simple LR" in Aho+Ullman terminology

do complete SLR DFA creation starting after initialization

dump the NFA

computing the "first" of the tail of a rule followed by an optional
terminal.

doesn't include NULLTOKEN
requires self.First to be computed

dump an item

this function completes the computation of an SLR DFA
by adding reduction states for each DFA state S containing
item   H > B.
which reduces rule H > B
for each token T in Follow of H.
if S already has a transition for T then there is a conflict!

assumes DFA and SLRNFA and Follow have been computed.

utility function -- returns true if an item is a final item