- Fixes a numerical issue where subsequent adjusted alphas are over-conservative (sometimes even zero) after a look in which there are many tied ranks among null test statistics (e.g. in network-based statistic where test stats are discrete, or if `n_permutations > 2**look_times[0]`).
- Suppresses `test_func`'s joblib text output during bootstrap power analysis with `bootstrap_predictive_power_1samp`, so users aren't bombarded with the statuses of thousands of permutation tests at once. (This issue only ever affected certain MNE versions.)