refactor(org-config): surface narrowing + sparse-tree under C-; O

Narrowing and sparse-tree commands existed in the `:bind' block
on `C-c'-style shortcuts but nothing in `cj/org-map' surfaced
them, so which-key never showed them and discoverability was
poor.

Add direct bindings under `C-; O', flat (no sub-prefixes for
narrow / sparse-tree).  Lowercase creates; capital of the same
letter cancels:

- `n' / `N'  narrow-to-subtree / widen
- `s' / `S'  match-sparse-tree / show-all
- `t' / `T'  show-todo-tree / show-all
- `>' / `<'  forward / backward sibling narrow (kept as-is)
- `R'        reveal-context (no lowercase pair -- `r' is the
             table-row sub-prefix)

Both `S' and `T' resolve to the same `org-show-all' command so
the mental model is just "capital cancels the lowercase I just
ran" without having to recall which letter the cancel actually
lives on.

Free up F2: the old `(<f2> . org-reveal)' binding in the org-mode
`:bind' block is now redundant with `C-; O R'.  Drop it; F2
becomes available for whatever wants it next.

Four new ERT assertions in `test-org-config-keymap-ownership.el'
lock the shape -- the old sparse-tree-submap test was rewritten
for the flat layout and the narrow-submap test became
narrow-bindings (also flat).

1 files changed, 25 insertions, 0 deletions

diff --git a/tests/test-org-config-keymap-ownership.el b/tests/test-org-config-keymap-ownership.el
index 7f08a172..729d497c 100644
--- a/tests/test-org-config-keymap-ownership.el
+++ b/tests/test-org-config-keymap-ownership.el
@@ -45,6 +45,31 @@ command (which moved to capital `C')."
   (should (eq (keymap-lookup cj/org-map "c i") #'org-table-insert-column))
   (should (eq (keymap-lookup cj/org-map "c d") #'org-table-delete-column)))
 
+(ert-deftest test-org-config-keymap-ownership-narrow-bindings ()
+  "Narrow / widen sit directly under `C-; O' (flat, no sub-prefix).
+Lowercase `n' narrows to the current subtree; capital `N' widens,
+matching the lowercase-creates / uppercase-cancels pattern shared
+with the sparse-tree sub-prefix.  Sibling-stepping is on `>' / `<'
+at the top level."
+  (should (eq (keymap-lookup cj/org-map "n") #'org-narrow-to-subtree))
+  (should (eq (keymap-lookup cj/org-map "N") #'widen))
+  (should (eq (keymap-lookup cj/org-map ">") #'cj/org-narrow-forward))
+  (should (eq (keymap-lookup cj/org-map "<") #'cj/org-narrow-backwards)))
+
+(ert-deftest test-org-config-keymap-ownership-sparse-tree-bindings ()
+  "Sparse-tree commands sit directly under `C-; O' (flat).
+Lowercase creates, capital of the same letter cancels: `s' /
+`S' for match-sparse-tree, `t' / `T' for show-todo-tree.  Both
+capitals resolve to `org-show-all' -- the user's mental model is
+\"capital cancels the lowercase I just ran\" without having to
+remember which letter the cancel actually lives on.  `R' is
+`org-reveal' (no lowercase pair -- `r' is the table-row sub-prefix)."
+  (should (eq (keymap-lookup cj/org-map "s") #'org-match-sparse-tree))
+  (should (eq (keymap-lookup cj/org-map "S") #'org-show-all))
+  (should (eq (keymap-lookup cj/org-map "t") #'org-show-todo-tree))
+  (should (eq (keymap-lookup cj/org-map "T") #'org-show-all))
+  (should (eq (keymap-lookup cj/org-map "R") #'org-reveal)))
+
 (ert-deftest test-org-config-keymap-ownership-regression-no-duplicate-org-keymap ()
   "The old duplicate `cj/org-keymap' binding should not exist."
   (should-not (boundp 'cj/org-keymap)))