Function fisher_two_tail_pvalue 

pub fn fisher_two_tail_pvalue(a: u32, b: u32, c: u32, d: u32) -> Option<f64>

Fisher’s exact two-tail p-value for the 2×2 contingency table

        | col 1 | col 2 |
--------+-------+-------+
row 1   |   a   |   b   |
row 2   |   c   |   d   |

Returns Some(p) with p in [0.0, 1.0], or None if the resulting hypergeometric distribution is degenerate (every row or column sum is zero — Fisher’s test is undefined).

Algorithm

Conditional on the row and column marginals, the count in the top-left cell is hypergeometric with parameters N = a+b+c+d, K = a+b (top row sum), and n = a+c (left column sum). The two-tail p-value is the sum of probabilities of all 2×2 tables with the same marginals whose probability under the hypergeometric model is at most that of the observed table — the standard Fisher exact convention (and what the prior fishers_exact crate computes).

All factorials are evaluated in log space via ln_factorial so the algorithm is numerically stable on u32 inputs — the marginal sums can reach a few million on a large monorepo coupling analysis, well beyond what f64::MAX can represent as a plain factorial.