Add smoke test and benchmarks

smoke links crt0.o + libc.a with -nostdlib -static and proves the whole chain runs without the dynamic linker; bench_strlen and bench_syscall measure the string core and raw syscall round-trip cost via make bench.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <[email protected]>

benchmarks/bench.h

@@ -0,0 +1,44 @@
+/*
+ * bench.h — tiny shared helpers for nulsl-libc benchmarks.
+ *
+ * Deliberately dependency-free beyond the libc itself: timing goes
+ * through a raw clock_gettime syscall, and numbers are printed with
+ * putchar/write because printf() is still a stub. If you write a new
+ * benchmark, include this and stay this lean.
+ */
+
+#ifndef _NULSL_BENCH_H
+#define _NULSL_BENCH_H
+
+#include <sys/syscall.h>
+#include <unistd.h>
+
+#define BENCH_CLOCK_MONOTONIC 1
+
+/* Monotonic time in nanoseconds, via raw clock_gettime(2). */
+static inline long long bench_now_ns(void)
+{
+    struct {
+        long tv_sec;
+        long tv_nsec;
+    } t;
+
+    syscall(SYS_clock_gettime, BENCH_CLOCK_MONOTONIC, &t);
+    return t.tv_sec * 1000000000LL + t.tv_nsec;
+}
+
+/* Print an unsigned integer plus a newline, without printf(). */
+static inline void bench_print_u64(unsigned long long v)
+{
+    char buf[24];
+    int i = sizeof buf;
+
+    buf[--i] = '\n';
+    do {
+        buf[--i] = (char)('0' + v % 10);
+        v /= 10;
+    } while (v);
+    write(STDOUT_FILENO, buf + i, sizeof buf - i);
+}
+
+#endif /* _NULSL_BENCH_H */