test+bins: flash grad-check, flash==composed, PyTorch parity, --flash flag

autograd: flash_attention_batched_bwd (dQ/dK/dV finite-diff, seq>tile)
+ flash_matches_composed_fwd. model/tests/flash.rs: flash==composed
on-vs-off (logits/loss/every param grad), fp32 + bf16. parity_dump:
XTRAIN_PARITY_FLASH dumps the flash path for the same parity.py oracle
(PyTorch SDPA parity at B>1). train + train_ddp get the --flash flag.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

crates/xtrain-distributed/src/bin/train_ddp.rs

@@ -89,6 +89,9 @@ fn main() {
     // rank checkpoints its own forward/backward; exact grads, lower peak activation
     // memory (lets dim1024 batch32 fit). Opt-in; default off.
     let recompute = args.iter().any(|a| a == "--recompute");
+    // Fused flash-attention (Phase T14): single fused SDPA kernel, online softmax,
+    // no materialized [bh,S,S] scores. Opt-in; default off keeps the composed path.
+    let flash = args.iter().any(|a| a == "--flash");
     let ckpt: Option<PathBuf> = args
         .iter()
         .position(|a| a == "--ckpt")
@@ -174,6 +177,9 @@ fn main() {
     if recompute {
         println!("activation recompute: ON (per-block gradient checkpointing)");
     }
+    if flash {
+        println!("flash-attention: ON (fused SDPA kernel, no materialized scores)");
+    }
     let results = launch(
         &devices,
         &train_corpus,
@@ -187,6 +193,9 @@ fn main() {
             if recompute {
                 m = m.with_recompute(true);
             }
+            if flash {
+                m = m.with_flash(true);
+            }
             m
         },
     );