reset result on apply

[funny-falcon]

It fixes duplicate production of previous rule application on non-producing entries:

rule_a = rule_b:b rule_c*:c { [b, c] }
rule_b = <"b"> { text }
rule_c = "c"

In example rule_c doesn't produce result, so result of rule_b is copied.

It first were mentioned in #41

After patch rule_c "produces" nil.

Fixes #41

[funny-falcon]

I don't know if it desired behavior or not.
Certainly result of previous rule should not be copied, but should nil be produced instead, or there should be more complex logic to avoid result at all?

[evanphx]

Thanks!