Author corrections without baking them into motion

Constant, ramp, and return offsets now append ordinary channel layers to a lane or cel in explicit owner frames. The inspector exposes the commands as one undoable transaction, shows conflicts, and offers removal or retry while preserving generated bases through regeneration.

Ordered-stack compatibility is shared by validation, conflict reporting, and regeneration, including adjacent replacement coverage. Cel-sheet gaps and headers select their lane, so commands cannot fall through to another column's stale selection.

437 tests, 5,804 assertions; both browser flows; 56 Django tests; optimized frontend build.

docs/lane-handoff.md

@@ -1,10 +1,11 @@
 # Lane and cel handoff
 
-Status (2026-09-30): the lane model is implemented through its commands and its
-first two views. Cels are ordinary nodes with their own playback clock, the
-timeline draws them as one row, the cel sheet draws frames down and lanes across,
-and both views issue the same commands. Correction layers evaluate and survive
-regeneration. What is missing is the commands that make a correction.
+Status (2026-09-30): the lane model is implemented through its commands, its
+first two views, and correction authoring. Cels are ordinary nodes with their
+own playback clock; the timeline draws them as one row and the cel sheet draws
+frames down and lanes across. Both views issue the same commands. Rotation and
+position corrections can be authored as Constant, Ramp, or Return motion on a
+lane or cel, survive regeneration, and expose conflicts for removal or retry.
 
 The commits beginning at `3d3c1bb` are the argument for the model and are worth
 reading before touching what they did — they are the design record, more than
@@ -87,18 +88,18 @@ decision, not a cleanup.
 
 ## Next steps, in order
 
-1. **The commands that make a correction** — Constant adjustment, Ramp, Return
-   motion over a selected range, per `lane-model.md`. The evaluator is done and
-   has no opinion about how a range or a motion shape is chosen, which is now a
-   view question. A panel also needs to offer `clip/conflicts` for resolution.
-   Note the one open question: a correction needs a stable `:id` from somewhere,
-   and cel ids come from the caller because this namespace is pure.
-2. **Slip source and retime.** Both have real design questions open and the doc
+The implemented correction slice and its remaining UI limits are recorded in
+[Correction authoring](correction-authoring-plan.md).
+
+1. **Slip source and retime.** Both have real design questions open and the doc
    says to refuse rather than approximate: retime needs a defined warp and
    interpolation behaviour, and is not moving keys whose numbers happen to fall
    inside a selection.
-3. **Deleting reused content.** Reference discovery exists (`node/sources`,
+2. **Deleting reused content.** Reference discovery exists (`node/sources`,
    `clip/places`, `clip/contains-symbol?`); the policy does not.
+3. **Displayed-range correction gestures.** The first correction panel asks for
+   explicit owner frames. Dragging a range in a retimed/nested view still needs
+   a proved mapping; do not make it snap through floors or loops.
 4. **Collaboration.** `lane-model.md` is explicit that one leaf per channel does
    NOT solve two people editing different keys of the same channel. No conflict
    policy exists for that.
@@ -126,6 +127,12 @@ decision, not a cleanup.
 - **Generated sampling applies to the base, not the hand correction.** Picture
   rate and pose selection may choose an earlier generated frame; correction
   support and values still read the node's current authored frame.
+- **Correction commands live in `domain/correction.cljs`.** IDs come from the
+  event caller; the pure command materializes default transform channels,
+  appends one layer, and validates the complete document. The inspector authors
+  rotation and position offsets in explicit owner frames. One Apply is one undo
+  step. `channel/reconcile` is the shared ordered-stack compatibility rule used
+  by validation, conflict reporting, and regeneration.
 - **Two test patterns worth copying.** `the-cursor-agrees-with-the-specification-in-any-frame-order`
   holds the optimized cursor to `value-at` in forward, backward and random order
   — add a case to it for any new channel shape. And `drawn` in `lane_test`
@@ -165,7 +172,7 @@ decision, not a cleanup.
 
 From `frontend/`:
 
-    npx shadow-cljs compile test && node out/node-tests.js     # 429 tests, 5,767 assertions
+    npx shadow-cljs compile test && node out/node-tests.js     # 437 tests, 5,804 assertions
     npx shadow-cljs compile app                                # the bundle Django serves
     npx shadow-cljs release app                                # then `compile app` again — see above