feat: full runtime redesign (v0.6)

Complete rewrite with improved architecture & correctness:
- src/runtime.rs: Simplified task scheduling with proper state transitions
- src/scheduler.rs: Decoupled from runtime, pure task queue logic
- src/io.rs, src/mutex.rs: Refactored for clarity & performance
- New actor model framework (src/actor.rs, src/context.rs)
- Channel primitives (src/channel.rs) & process IDs (src/pid.rs)
- Preemption framework (src/preempt.rs) for fair timeslicing
- Expanded benchmarks & tests (multi_scheduler, primes, runtime)

tests/channel.rs

@@ -0,0 +1,110 @@
+//! Channel tests. These run under the scheduler because `recv()` needs to
+//! be able to park, which requires a live runtime.
+
+use smarm::{channel, run, spawn};
+use std::cell::Cell;
+
+thread_local! {
+    static OUT: Cell<i64> = const { Cell::new(0) };
+}
+
+#[test]
+fn send_then_recv_same_actor() {
+    OUT.with(|c| c.set(0));
+    run(|| {
+        let (tx, rx) = channel::<i64>();
+        tx.send(42).unwrap();
+        let v = rx.recv().unwrap();
+        OUT.with(|c| c.set(v));
+    });
+    assert_eq!(OUT.with(|c| c.get()), 42);
+}
+
+#[test]
+fn recv_parks_until_send_from_other_actor() {
+    OUT.with(|c| c.set(0));
+    run(|| {
+        let (tx, rx) = channel::<i64>();
+        let h = spawn(move || {
+            // This actor blocks on an empty channel.
+            let v = rx.recv().unwrap();
+            OUT.with(|c| c.set(v));
+        });
+        // Parent runs, then yields to let the child block,
+        // then sends, then joins.
+        smarm::yield_now();
+        tx.send(7).unwrap();
+        h.join().unwrap();
+    });
+    assert_eq!(OUT.with(|c| c.get()), 7);
+}
+
+#[test]
+fn multiple_messages_arrive_in_order() {
+    let captured: std::sync::Arc<std::sync::Mutex<Vec<i64>>> =
+        std::sync::Arc::new(std::sync::Mutex::new(Vec::new()));
+    let cap2 = captured.clone();
+
+    run(move || {
+        let (tx, rx) = channel::<i64>();
+        let h = spawn(move || {
+            for _ in 0..3 {
+                let v = rx.recv().unwrap();
+                cap2.lock().unwrap().push(v);
+            }
+        });
+        for v in 1..=3i64 {
+            tx.send(v).unwrap();
+        }
+        h.join().unwrap();
+    });
+
+    assert_eq!(*captured.lock().unwrap(), vec![1, 2, 3]);
+}
+
+#[test]
+fn cloned_senders_both_deliver() {
+    let captured: std::sync::Arc<std::sync::Mutex<Vec<i64>>> =
+        std::sync::Arc::new(std::sync::Mutex::new(Vec::new()));
+    let cap2 = captured.clone();
+
+    run(move || {
+        let (tx, rx) = channel::<i64>();
+        let tx2 = tx.clone();
+        let h = spawn(move || {
+            for _ in 0..2 {
+                let v = rx.recv().unwrap();
+                cap2.lock().unwrap().push(v);
+            }
+        });
+        tx.send(10).unwrap();
+        tx2.send(20).unwrap();
+        h.join().unwrap();
+    });
+
+    let mut got = captured.lock().unwrap().clone();
+    got.sort();
+    assert_eq!(got, vec![10, 20]);
+}
+
+#[test]
+fn recv_returns_err_when_all_senders_dropped() {
+    let saw_err: std::sync::Arc<std::sync::atomic::AtomicBool> =
+        std::sync::Arc::new(std::sync::atomic::AtomicBool::new(false));
+    let saw_err2 = saw_err.clone();
+
+    run(move || {
+        let (tx, rx) = channel::<i64>();
+        let h = spawn(move || {
+            // Receiver waits; no message will ever come.
+            if rx.recv().is_err() {
+                saw_err2.store(true, std::sync::atomic::Ordering::SeqCst);
+            }
+        });
+        smarm::yield_now();
+        drop(tx); // last sender gone; rx.recv must return Err.
+        h.join().unwrap();
+    });
+
+    assert!(saw_err.load(std::sync::atomic::Ordering::SeqCst));
+}