docs(ltx): correct the refine default comment and stale file references

The refine-default comment claimed Condition variants are excluded from two-stage
refine, but supports_two_stage_refine has no such carve-out and ltx_process builds
a second condition set so conditioning survives the upsample. Two file references
had also drifted: ltx_process.py:179 moved, and the offload hook keying now lives
in sd_offload_balanced. Name the symbols instead of the line numbers.

modules/ltx/ltx_process.py

@@ -418,7 +418,7 @@ def run(model: str, *,
                 t2 = time.time()
                 # silent=True everywhere: per-module stats were already dumped during the load-time
                 # balanced_offload pass. Upsample/refine boundaries force a rebuild because the global
-                # offload_hook_instance is keyed on checkpoint_name (sd_offload.py:488), but re-logging
+                # offload_hook_instance is keyed on checkpoint_name (sd_offload_balanced), but re-logging
                 # the same inventory adds noise without information.
                 shared.sd_model = sd_models.apply_balanced_offload(shared.sd_model, silent=True)
                 devices.torch_gc(force=True, reason='ltx:base')

modules/ltx/ltx_ui.py

@@ -33,9 +33,8 @@ def _model_change(model_name: str):
         )
     # 2.x refine runs fixed canonical schedules; refine_strength only feeds 0.9.x LTXConditionPipeline.
     refine_strength_interactive = caps.family == '0.9'
-    # Default Refine on for any 2.x variant whose refine path expects upsampled latents (Dev and
-    # Distilled T2V/I2V). auto_refine_upsample at ltx_process.py:179 couples the stages once Refine
-    # is on. Condition variants are excluded by supports_two_stage_refine.
+    # Default Refine on for every 2.x row: ltx_process couples it to an implicit 2x upsample,
+    # since both refine paths expect upsampled latents and same-resolution refine oversaturates.
     refine_default = caps.supports_two_stage_refine
     auto_duration_update =gr.update(visible=True) if caps.supports_auto_duration else gr.update(visible=False, value=False)
     return (