Release 2.2.0 — pyeeg.stats module (Issue #14)

New features:
- pyeeg.stats: nonparametric statistical inference for TRF analysis
  - permutation_test_trf (circular-shift null, stat=zscore/t/coef/perm_norm)
  - cluster_based_permutation_test (Maris & Oostenveld 2007)
  - bootstrap_ci_trf (paired block bootstrap, boundary drop)
  - jackknife_se_trf (leave-one-epoch-out SE/CI, any solver)
  - cross_subject_consistency (descriptive reliability)
  - group_level_test (sign-flip, H0: population mean = 0)
  - Spectral edge fade for autocorrelated stimuli
  - No MNE dependency

Fixes:
- TRFEstimator.copy() now preserves all constructor kwargs

451 tests passing.