feat(input): #83 Input Manager + auto-commit the device layer in the frame loop

Fixes the papercut: the generated frame loop called rt_poll() (feeding only
Input.key) but never input_poll(), so Input.active/key_down/mouse/pad read
empty unless the game called Input.poll() by hand. The loop now calls
input_poll() when the game uses any Input runtime method — committing the
held-key/mouse/gamepad state, and record/replay — and stores its return as the
frame key so Input.key still works. A game using no Input runtime keeps the
plain rt_poll path, byte-identical.

Adds the Input-Manager API: Input.action(name,key) ships a default binding
(kept if already bound, so a rebind/loaded map isn't clobbered),
Input.bind_pad(name,button) makes an action device-agnostic (keyboard OR pad),
and Input.active/just_pressed/just_released read the multi-key device layer
with clean on-press/on-release edges (deterministic, dispatch-free — a handler
polls the edge; a replay fires identically).

Examples input_manager + input_auto, 5 docs pages. Full suite 107/0, goldens
byte-identical, fixpoint holds.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/language/input/input-just_released.md

@@ -0,0 +1,23 @@
+---
+id: input-just_released
+name: Input.just_released
+category: input
+kind: namespace-method
+tokens: Input.just_released
+sig: Input.just_released(action: str) -> bool
+tip: Did a named action go inactive this frame (the on-release edge)?
+order: 35
+ns: Input
+member: just_released
+---
+
+Returns whether a named action went inactive this frame — not active now, active last frame. The on-release edge over the device layer (keyboard or pad), symmetric with <a href="input-just_pressed.html"><code>Input.just_pressed</code></a>. Use it for release-triggered actions (variable-height jump cut-off, charge-and-release). Deterministic, so a replay fires the same edges.
+
+```ludic
+program Release {
+  entry {
+    Input.action("charge", ' ')
+    if Input.just_released("charge") { print(1) }
+  }
+}
+```