lightningd: fix crash on rapid reconnect.

Happens occasionally when running
`tests/test_connection.py::test_mutual_reconnect_race` (which is too
flaky to add, without more fixes):


```
lightningd: lightningd/peer_control.c:1252: peer_active: Assertion `!channel->owner' failed.
lightningd: FATAL SIGNAL 6 (version v0.11.0.1-38-g4f167da)
0x5594a41f8f45 send_backtrace
	common/daemon.c:33
0x5594a41f8fef crashdump
	common/daemon.c:46
0x7f7cb585c08f ???
	/build/glibc-SzIz7B/glibc-2.31/signal/../sysdeps/unix/sysv/linux/x86_64/sigaction.c:0
0x7f7cb585c00b __GI_raise
	../sysdeps/unix/sysv/linux/raise.c:51
0x7f7cb583b858 __GI_abort
	/build/glibc-SzIz7B/glibc-2.31/stdlib/abort.c:79
0x7f7cb583b728 __assert_fail_base
	/build/glibc-SzIz7B/glibc-2.31/assert/assert.c:92
0x7f7cb584cfd5 __GI___assert_fail
	/build/glibc-SzIz7B/glibc-2.31/assert/assert.c:101
0x5594a41b45ca peer_active
	lightningd/peer_control.c:1252
0x5594a418794c connectd_msg
	lightningd/connect_control.c:457
0x5594a41cd457 sd_msg_read
	lightningd/subd.c:556
0x5594a41ccbe5 read_fds
	lightningd/subd.c:357
0x5594a4269fc2 next_plan
	ccan/ccan/io/io.c:59
0x5594a426abca do_plan
	ccan/ccan/io/io.c:407
0x5594a426ac0c io_ready
	ccan/ccan/io/io.c:417
0x5594a426ceff io_loop
	ccan/ccan/io/poll.c:453
0x5594a41930d9 io_loop_with_timers
	lightningd/io_loop_with_timers.c:22
0x5594a4199293 main
	lightningd/lightningd.c:1181
0x7f7cb583d082 __libc_start_main
	../csu/libc-start.c:308
0x5594a416e15d ???
	???:0
0xffffffffffffffff ???
	???:0
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

lightningd/peer_control.c

@@ -1252,7 +1252,8 @@ void peer_active(struct lightningd *ld, const u8 *msg, int fd)
 		case CHANNELD_NORMAL:
 		case CHANNELD_SHUTTING_DOWN:
 		case CLOSINGD_SIGEXCHANGE:
-			assert(!channel->owner);
+			/* Maybe old owner was too slow exiting? */
+			tal_free(channel->owner);
 			peer_start_channeld(channel,
 					    peer_fd,
 					    NULL, true,