migration/ram: fix memleaks in multifd_new_send_channel_async

When error happen in multifd_new_send_channel_async, 'sioc' will not be used
to create the multifd_send_thread. Let's free it to avoid a memleak. And also
do error_free after migrate_set_error() to avoid another leak in the same place.

The leak stack:
Direct leak of 2160 byte(s) in 6 object(s) allocated from:
    #0 0xfffdd97fe754 in malloc (/lib64/libasan.so.4+0xee754)
    #1 0xfffdd85a8b48 in g_malloc (/lib64/libglib-2.0.so.0+0x58b48)
    #2 0xaaadfc4e2b10 in object_new_with_type qom/object.c:634
    #3 0xaaadfc619468 in qio_channel_socket_new io/channel-socket.c:56
    #4 0xaaadfc3d3e74 in socket_send_channel_create migration/socket.c:37
    #5 0xaaadfbaed6f4 in multifd_save_setup /usr/src/debug/qemu-4.1.0-4_asan.aarch64/migration/ram.c:1255
    #6 0xaaadfc3d2f78 in migrate_fd_connect migration/migration.c:3359
    #7 0xaaadfc3d6240 in migration_channel_connect migration/channel.c:101
    #8 0xaaadfc3d3590 in socket_outgoing_migration migration/socket.c:108
    #9 0xaaadfc625a64 in qio_task_complete io/task.c:195
    #10 0xaaadfc625ed0 in qio_task_thread_result io/task.c:111
    #11 0xfffdd859edec  (/lib64/libglib-2.0.so.0+0x4edec)
    #12 0xfffdd85a2a78 in g_main_context_dispatch (/lib64/libglib-2.0.so.0+0x52a78)
    #13 0xaaadfc6d3b84 in glib_pollfds_poll util/main-loop.c:218
    #14 0xaaadfc6d3b84 in os_host_main_loop_wait util/main-loop.c:241
    #15 0xaaadfc6d3b84 in main_loop_wait util/main-loop.c:517
    #16 0xaaadfbf9206c in main_loop /usr/src/debug/qemu-4.1.0-4_asan.aarch64/vl.c:1791
    #17 0xaaadfba1b124 in main /usr/src/debug/qemu-4.1.0-4_asan.aarch64/vl.c:4473
    #18 0xfffdd7833f5c in __libc_start_main (/lib64/libc.so.6+0x23f5c)
    #19 0xaaadfba26360  (/usr/libexec/qemu-kvm+0x886360)

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>