fix(h3): JoyAI-Echo is a LoRA technique, not just a frame-index conflict

The prior wording made the frame-index/anchor-eviction issue read as the
reason JoyAI is blocked on H3. It isn't: no JoyAI-Echo LoRA exists for H3, so
it's inaccessible by design regardless of interior pins. The frame-index issue
is real but secondary — it's what would also go wrong if someone forced it on.

movie_editor/frontend/pipeline_caps.js

@@ -137,10 +137,12 @@
       + "condition row rather than an appended frame, so that tail is always empty. "
       + "(alg_anchor is NOT hidden — a continuation scene does carry real latent frames.)",
     joyai_memory:
-      "JoyAI-Echo places memory frame i at sequence position i. H3's packed layout pins only "
-      + "the FIRST or LAST frame, so every frame past the first is refused — and the one that "
-      + "lands replaces the scene's i2v anchor, because pins are keyed by frame index. This "
-      + "one is not merely inert on H3, it is harmful, so the sampler forces it off.",
+      "JoyAI-Echo is a LoRA-driven technique: the base weights were never trained to read the "
+      + "injected memory frames as memory, the LoRA is what teaches that, and no JoyAI-Echo "
+      + "LoRA exists for H3. There is nothing to load, so this is inaccessible on H3 by "
+      + "design, not merely unwired. It would also actively hurt output if forced on: the "
+      + "bank places memory frame i at sequence position i, and writing index 0 twice "
+      + "replaces whatever was there — evicting the scene's own i2v anchor.",
   };
 
   // Sub-settings of the two groups above: they configure machinery that cannot run here.

samplers.py

@@ -7326,11 +7326,13 @@ class FunPackLTXAVSceneChainSampler:
                              "tail is always 0 and there is nothing to blur. alg_anchor still "
                              "works — a continuation scene does carry real latent frames)")
             if joyai_memory:
-                _dead.append("joyai_memory / JoyAI-Echo (the memory bank places frame i at "
-                             "sequence position i, and H3's packed layout pins only the FIRST "
-                             "or LAST frame — so every frame past the first is refused, and "
-                             "the one that lands REPLACES the scene's i2v anchor, since pins "
-                             "are keyed by frame index. Not inert on H3: harmful)")
+                _dead.append("joyai_memory / JoyAI-Echo (a LoRA-driven technique — the base "
+                             "weights were never trained to read the injected memory frames "
+                             "as memory, the LoRA teaches that, and no JoyAI-Echo LoRA exists "
+                             "for H3. Inaccessible by design, not merely unwired. It would "
+                             "also actively hurt output if forced on: the bank places memory "
+                             "frame i at sequence position i, and writing index 0 twice "
+                             "REPLACES whatever was there, evicting the scene's i2v anchor)")
             # Gated on joyai_audio_memory, not on the value alone: v2a_grad_scale is that
             # feature's coupling knob and is never installed without it (see the call site),
             # so reporting a left-over value as an H3 limitation blames the model for a knob