feat(synth): 7 new SCM methods — Bayesian, BSTS/CausalImpact, Penalized (A&L 2021), FDID, Cluster, Sparse, Kernel

Expand synthetic control module from 13 to 20 methods, making StatsPAI
the most comprehensive SCM toolkit in any language (Python, R, or Stata).

New modules (5,009 LOC):
- bayesian.py: Full Bayesian SCM with MCMC posterior inference
- bsts.py: Bayesian Structural Time Series / CausalImpact (Brodersen et al. 2015)
- penscm.py: Penalized SCM with pairwise discrepancy (Abadie & L'Hour, JASA 2021)
- fdid.py: Forward DID with optimal donor selection (Li 2024)
- cluster.py: Cluster SCM with donor grouping (kmeans/spectral/hierarchical)
- sparse.py: Sparse SCM with L1 penalties (Amjad, Shah & Shen 2018)
- kernel.py: Kernel-based nonlinear SCM (RBF, polynomial, Laplacian)

Integration:
- synth() dispatcher routes all 20 methods via unified API
- __init__.py exports 9 new public functions
- compare.py method registry updated for benchmarking
- 50 new tests (119 total synth tests, all passing)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: brycewang-stanford

src/statspai/__init__.py

@@ -60,6 +60,10 @@
     discos, qqsynth, discos_test, discos_plot, stochastic_dominance,
     synth_loo, synth_time_placebo, synth_donor_sensitivity,
     synth_rmspe_filter, synth_sensitivity, synth_sensitivity_plot,
+    synth_power, synth_mde, synth_power_plot,
+    synth_compare, synth_recommend, SynthComparison,
+    synth_report, synth_report_to_file,
+    german_reunification, basque_terrorism, california_tobacco,
     synthdid_estimate, sc_estimate, did_estimate,
     synthdid_placebo, synthdid_plot, synthdid_units_plot, synthdid_rmse_plot,
     california_prop99,

src/statspai/synth/__init__.py

@@ -3,7 +3,7 @@
 
 Unified entry point: ``synth(method=...)`` dispatches to all variants.
 
-Variants (13 methods)
+Variants (20 methods)
 ---------------------
 - **classic** — Abadie, Diamond & Hainmueller (2010)
 - **penalized / ridge** — Ridge-penalised SCM
@@ -17,13 +17,22 @@
 - **discos / distributional** — Gunsilius (2023)
 - **multi_outcome** — Sun (2023)
 - **scpi / prediction_interval** — Cattaneo, Feng & Titiunik (2021)
+- **bayesian** — Bayesian SCM with MCMC posterior (Vives & Martinez 2024)
+- **bsts / causal_impact** — Bayesian Structural Time Series (Brodersen et al. 2015)
+- **penscm / abadie_lhour** — Penalized SCM with pairwise discrepancy (Abadie & L'Hour 2021)
+- **fdid / forward_did** — Forward DID with optimal donor selection (Li 2024)
+- **cluster** — Cluster SCM with donor grouping (Rho 2024)
+- **sparse / lasso** — Sparse SCM with L1 penalties (Amjad, Shah & Shen 2018)
+- **kernel / kernel_ridge** — Kernel-based nonlinear SCM
 
 Inference
 ---------
 - **placebo** — in-space permutation (default)
 - **conformal** — Chernozhukov, Wüthrich & Zhu (2021)
 - **bootstrap / jackknife** — for SDID
 - **prediction intervals** — Cattaneo et al. (2021)
+- **bayesian posterior** — full posterior credible intervals (Bayesian SCM)
+- **bsts posterior** — Bayesian structural time series uncertainty
 
 Diagnostics
 -----------
@@ -54,6 +63,36 @@
 # Distributional Synthetic Controls
 from .discos import discos, qqsynth, discos_test, discos_plot, stochastic_dominance
 
+# Bayesian SCM
+from .bayesian import bayesian_synth
+
+# BSTS / CausalImpact
+from .bsts import causal_impact, bsts_synth
+
+# Penalized SCM (Abadie & L'Hour 2021)
+from .penscm import penalized_synth
+
+# Forward DID
+from .fdid import fdid
+
+# Cluster SCM
+from .cluster import cluster_synth
+
+# Sparse SCM
+from .sparse import sparse_synth
+
+# Kernel / Nonlinear SCM
+from .kernel import kernel_synth, kernel_ridge_synth
+
+# Multi-method comparison & auto-recommendation
+from .compare import synth_compare, synth_recommend, SynthComparison
+
+# Power analysis & sample size planning
+from .power import synth_power, synth_mde, synth_power_plot
+
+# Report generator
+from .report import synth_report, synth_report_to_file
+
 # Sensitivity & robustness diagnostics
 from .sensitivity import (
     synth_loo,
@@ -77,11 +116,18 @@
     california_prop99,
 )
 
+# Canonical SCM datasets
+from .datasets import (
+    german_reunification,
+    basque_terrorism,
+    california_tobacco,
+)
+
 __all__ = [
     # Unified entry point
     'synth',
     'SyntheticControl',
-    # Variant shortcuts
+    # Variant shortcuts (original 13)
     'demeaned_synth',
     'robust_synth',
     'gsynth',
@@ -100,12 +146,33 @@
     'discos_test',
     'discos_plot',
     'stochastic_dominance',
+    # New methods (7 additions)
+    'bayesian_synth',
+    'causal_impact',
+    'bsts_synth',
+    'penalized_synth',
+    'fdid',
+    'cluster_synth',
+    'sparse_synth',
+    'kernel_synth',
+    'kernel_ridge_synth',
     # SDID framework
     'sdid',
     'synthdid_estimate',
     'sc_estimate',
     'did_estimate',
     'synthdid_placebo',
+    # Multi-method comparison
+    'synth_compare',
+    'synth_recommend',
+    'SynthComparison',
+    # Report generator
+    'synth_report',
+    'synth_report_to_file',
+    # Power analysis
+    'synth_power',
+    'synth_mde',
+    'synth_power_plot',
     # Sensitivity & robustness
     'synth_loo',
     'synth_time_placebo',
@@ -120,4 +187,7 @@
     'synthdid_rmse_plot',
     # Data
     'california_prop99',
+    'german_reunification',
+    'basque_terrorism',
+    'california_tobacco',
 ]