LLVM: fix lowering of structs with underaligned fields

When lowering a struct type to an LLVM struct type, keep track of
whether there are any underaligned fields. If so, then make it a packed
llvm struct. This works because we already insert manual padding bytes
regardless.

We could unconditionally use an LLVM packed struct; the reason we bother
checking for underaligned fields is that it is a conservative choice, in
case LLVM handles packed structs less optimally. A future improvement
could simplify this code by unconditionally using packed LLVM structs
and then make sure measure perf is unaffected.

closes #12190

lib/std/os/linux.zig

@@ -3236,7 +3236,10 @@ pub const epoll_event = switch (builtin.zig_backend) {
     },
     else => extern struct {
         events: u32,
-        data: epoll_data align(4),
+        data: epoll_data align(switch (native_arch) {
+            .x86_64 => 4,
+            else => @alignOf(epoll_data),
+        }),
     },
 };
 

lib/std/x/os/io.zig

@@ -117,7 +117,6 @@ pub const Reactor = struct {
 };
 
 test "reactor/linux: drive async tcp client/listener pair" {
-    if (@import("builtin").zig_backend != .stage1) return error.SkipZigTest;
     if (native_os.tag != .linux) return error.SkipZigTest;
 
     const ip = std.x.net.ip;