Correct how the work is sequenced, and record #15 opened

docs/IRIS_EXTRACTION_HANDOFF.md

@@ -9,17 +9,17 @@ Canonical `main` is **`32b1038`** (#14, SetSize). Twelve slices are in. One
 pull request is open and one branch is pushed but unopened:
 
 - **#12** `split/12-pointer-routing`, worktree `/home/bob/repos/iris-pr12`,
-  head `f3fd941`. Reworked after the owner's review rejected the design: a
-  layer, not a widget, is what consumes input. `CursorSenses::consumes`
-  decides only whether a layer stops the input reaching the one below, taking
-  nothing out of the cursor, and senses on one layer never block each other.
-  Five tests; two fail on `main`. One question is out to her in the reply:
-  whether a widget the cursor has left should still be handed a press.
-- **`split/15-harness`**, worktree `/home/bob/repos/iris-pr15`, head
-  `d5efdd2`, pushed with no pull request opened. `Tasks` takes an
-  `Arc<dyn WakeTaskQueue>` instead of an `Arc<Window>`, so `DefaultRsc` builds
-  without one, and `iris::harness` drives a ui with no surface. Six tests in
-  `tests/harness.rs`. Open it when #12 clears.
+  head `e53ce58`. Reworked twice after review: a layer, not a widget, is what
+  consumes input. `CursorSenses::consumes` decides only whether a layer stops
+  the input reaching the one below, and senses on one layer never block each
+  other. The vocabulary is `position_only`, not "momentary". Five tests; two
+  fail on `main`. **One question is out to her**: whether to replace
+  `registered` with a `consumed: &Cell<bool>` on `CursorData` that
+  `should_run` sets, which also removes the double match.
+- **#15** `split/15-harness`, worktree `/home/bob/repos/iris-pr15`. `Tasks`
+  takes an `Arc<dyn WakeTaskQueue>` instead of an `Arc<Window>`, so
+  `DefaultRsc` builds without one, and `iris::harness` drives a ui with no
+  surface. Six tests in `tests/harness.rs`. Disjoint from #12.
 
 Check for a review before starting anything, and read the newest
 `submitted_at` rather than the first result:
@@ -73,12 +73,16 @@ Other standing instructions from the owner:
 - Pull Iris out even if the Rust application switchover is not accepted. No
   app, session, transcript, setup or server concepts in Iris; the dependency
   runs one way from `app/` to Iris.
-- **Small, coherent PRs, one at a time.** The original extraction PR was too
-  large to review. A slice may be redone rather than transplanted, and need not
-  remove every old feature. Do not open several at once to stay busy during a
-  review: on 2026-09-13 the owner asked for the work to go serially, most
-  fundamental first, so that code is written once against the version of the
-  framework that will actually exist.
+- **Small, coherent PRs.** The original extraction PR was too large to review.
+  A slice may be redone rather than transplanted, and need not remove every old
+  feature. Non-conflicting pull requests may be open at once — disjoint path
+  sets, each branched from current `upstream/main` rather than stacked. The
+  owner reviews small ones as they arrive and only avoids having two *large*
+  ones in flight, which is one more reason to keep a slice small.
+- **Order by dependency, largest reach first.** On 2026-09-13: *"do the large
+  reaching framework changes first so less has to be redone."* Pick the next
+  slice by how much sits on top of it, not by what is nearest ready, so each
+  piece of code is written once against the framework that will exist.
 - Do not recreate an `ai` branch in canonical Iris; the fork is the boundary.
 - **Never rewrite a pushed branch.** Follow review with additive commits, and
   merge `upstream/main` in rather than rebasing when a branch falls behind.