compute exact p-value for small null configs

When num_pos and num_total is relatively small, e.g. (5 replicates, 20 negcon), total combinations is 4845. Here, instead of building `null_size=10,000` random ranks, p_val should be estimated by computing the 4,845 average precision scores.

In general, if `math.comb(num_total, num_pos)

Start by locating the code that uses num_pos, num_total, null_size, and p_val, then inspect how average precision scores are generated for random null ranks. Use math.comb(num_total, num_pos) as the branch condition; done means configurations below null_size use all combinations and produce the exact p-value, with existing behavior preserved otherwise.

Written by the indexing model from the issue text.