I calculated 0.006% from a software. In notes they mentioned similarity of standard normal distribution (as df is large) but even it gives 0.003%(not accurate as t-distribution ) So I thinkk 0.013% is erratum.
I can see at 0.001 percentage 4.2649 for a standard normal distribution. I think they interpolated it for 4.003. After posting here, it stuck me.
For any good interpretation null hypothesis is rejected. So no worries! But point is either ActEd wrote 2*(p-value(exact) ) or it is 0.003(as per table) (in notes written 0.013)