fix(prithvi): drop .view() on lists; enrich warmup traceback

Diag output from the previous deploy:

warm/prithvi: AttributeError: 'list' object has no attribute 'view'

The yaml-instantiated GenericNonGeoSegmentationDataModule stores
`means` and `stds` as Python lists (verified against the v2
config). The previous patch chained `_old.means.view(-1)
.detach().clone()` BEFORE handing them to _DictNormalize, which
crashed on the .view() call before reaching the conversion logic
inside the class.

Fix: pass `_old.means` and `_old.stds` to _DictNormalize raw.
The class's `_torch.as_tensor(mean).view(-1, 1, 1).float()` handles
list-to-tensor conversion internally — same end state, no
precondition that assumes the inputs are already tensors.

Also: lifespan warmup now stores a 5-line traceback tail in
_LAST_ERR so the next regression surfaces with a real stack
location through /healthz and /v1/diag.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

app/flood_layers/prithvi_live.py

@@ -233,10 +233,14 @@ def _ensure_model():
                                 std = self.std.to(sample.device)
                                 return (sample - mean) / std
 
-                        m.datamodule.aug = _DictNormalize(
-                            _old.means.view(-1).detach().clone(),
-                            _old.stds.view(-1).detach().clone(),
-                        )
+                        # `_old.means` / `_old.stds` come from the
+                        # yaml as Python lists — calling `.view()` on
+                        # them is what tripped the original
+                        # `'list' object has no attribute 'view'`.
+                        # _DictNormalize handles the conversion via
+                        # torch.as_tensor internally; just pass the
+                        # raw values whatever their type.
+                        m.datamodule.aug = _DictNormalize(_old.means, _old.stds)
                         log.info("prithvi_live: patched v2 datamodule transforms "
                                  "for IBM inference.py compat (dict-aware Normalize)")
                 else:

services/riprap-models/main.py

@@ -173,10 +173,13 @@ def _load_prithvi():
                         std = self.std.to(sample.device)
                         return (sample - mean) / std
 
-                m.datamodule.aug = _DictNormalize(
-                    _old.means.view(-1).detach().clone(),
-                    _old.stds.view(-1).detach().clone(),
-                )
+                # `_old.means` / `_old.stds` come from the yaml as
+                # Python lists — calling `.view()` on them is what
+                # tripped the original `'list' object has no attribute
+                # 'view'`. _DictNormalize handles the conversion via
+                # torch.as_tensor internally; just pass the raw values
+                # whatever their type.
+                m.datamodule.aug = _DictNormalize(_old.means, _old.stds)
                 log.info("prithvi: patched v2 datamodule transforms "
                          "for IBM inference.py compat (dict-aware Normalize)")
         else:
@@ -803,10 +806,15 @@ async def lifespan(_app: FastAPI):
                 fn()
                 log.info("startup %s ok", stage)
             except Exception as e:  # noqa: BLE001
+                import traceback
+                tb = traceback.format_exc()
                 log.exception("startup %s failed: %s", stage, e)
-                _LAST_ERR[stage] = {"ok": False,
-                                     "err": f"{type(e).__name__}: {e}",
-                                     "stage": stage}
+                _LAST_ERR[stage] = {
+                    "ok": False,
+                    "err": f"{type(e).__name__}: {e}",
+                    "stage": stage,
+                    "traceback_tail": tb.splitlines()[-5:],
+                }
     yield
     log.info("riprap-models stopping")