Add full quantile stats and post-hoc calibration

report.md

@@ -145,6 +145,7 @@ Key steps:
 - Streaming mean/std/min/max + int-like detection
 - Optional **log1p transform** for heavy-tailed continuous columns
 - Optional **quantile transform** (TabDDPM-style) for continuous columns (skips extra standardization)
+- Optional **post-hoc quantile calibration** to align 1D CDFs after sampling
 - Discrete vocab + most frequent token
 - Windowed batching with **shuffle buffer**
 
@@ -161,7 +162,8 @@ Export process:
 - Output: `trend + residual`
 - De-normalize continuous values
 - Inverse quantile transform (if enabled; no extra de-standardization)
-- Bound to observed min/max (clamp or sigmoid mapping)
+- Optional post-hoc quantile calibration (if enabled)
+- Bound to observed min/max (clamp / sigmoid / soft_tanh / none)
 - Restore discrete tokens from vocab
 - Write to CSV