Assuming all ten coins are of the same mint, we find P(average of head-counts for the 10 coins is extreme). To do that we need to know the distribution of this average. It can be calculated explicitly with a bit of probability. In fact, the sum of (independent) Binomials with the same probability of success is another Binomial.
And again, if the p-value is too small, we reject the null hypothesis of P(heads)=1/2.