G8: a notice leaves quicker than it arrives

The notice chip's motion was mostly there (a drop easing out on the
Motion flavour, a dissolve easing in, Lift's shadow in tier 4 at its
fade). Its dissolve took 300 ms against a 180 ms fall, the wrong way
round for §8.1 (leaving quicker than arriving, 100 to 160 ms):
MessageDissolve now defaults to 150. The field comments had the two
easings swapped; the stack test pins a longer dissolve for the case it
checks. docs/config.md and docs/effects.md say so.

Tests: both timings in their bands, leaving quicker. Gates as G7 (no
golden moved). Feel review: .scratch/render/notices/.

1 files changed, 9 insertions, 0 deletions

diff --git a/src/Messages.zig b/src/Messages.zig
index 7d284e6b..36cd44da 100644
--- a/src/Messages.zig
+++ b/src/Messages.zig
@@ -768,6 +768,8 @@ test "messages stack, oldest on top, each on its own timing" {
     const pane = p.panes[p.active].?;
     try std.testing.expect(p.executeBuiltinLine(p.active, "Verbose off"));
     try std.testing.expect(p.executeBuiltinLine(p.active, "MessageLinger 160"));
+    // A dissolve longer than a fall, so the second is caught mid-way.
+    try std.testing.expect(p.executeBuiltinLine(p.active, "MessageDissolve 300"));
     // `Verbose off` announced itself before it took effect.
     pane.msg_len = 0;
     pane.msg_life = .{};
@@ -968,3 +970,10 @@ test "an idle core wants no wake and a lingering message sleeps to its linger's
     p.advance(long_due);
     try std.testing.expectEqual(Pane.MessagePhase.leaving, pane.msg_life.phase);
 }
+
+test "a notice leaves quicker than it arrives, each within its §8.1 band" {
+    const settings: config.Runtime = .{};
+    try std.testing.expect(settings.message_dissolve_ms < settings.message_fall_ms);
+    try std.testing.expect(settings.message_fall_ms >= 150 and settings.message_fall_ms <= 220);
+    try std.testing.expect(settings.message_dissolve_ms >= 100 and settings.message_dissolve_ms <= 160);
+}