It's foolish to ban passing NULL, 0 to memcpy, memset et al, but
it's been done.  At high level of optimization, GCC assumes this doesn't
happen, and yep, assumes "if (ctx)" inside tal_free() must be true.

So when a psbt is NULL, and psbt_get_bytes returns NULL, a crash ensues:

```
lightningd: FATAL SIGNAL 6 (version v0.12.0rc2-6-g47efa5d-modded)
0x5557dfc42fef send_backtrace
	common/daemon.c:33
0x5557dfc42fef crashdump
	common/daemon.c:46
0x7fe93ef5851f ???
	./signal/../sysdeps/unix/sysv/linux/x86_64/libc_sigaction.c:0
0x7fe93efaca7c __pthread_kill_implementation
	./nptl/pthread_kill.c:44
0x7fe93efaca7c __pthread_kill_internal
	./nptl/pthread_kill.c:78
0x7fe93efaca7c __GI___pthread_kill
	./nptl/pthread_kill.c:89
0x7fe93ef58475 __GI_raise
	../sysdeps/posix/raise.c:26
0x7fe93ef3e7f2 __GI_abort
	./stdlib/abort.c:79
0x5557dfbb0c28 call_error
	ccan/ccan/tal/tal.c:93
0x5557dfbb0c34 check_bounds
	ccan/ccan/tal/tal.c:165
0x5557dfbb0c34 to_tal_hdr
	ccan/ccan/tal/tal.c:178
0x5557dfc7a1d3 tal_free
	ccan/ccan/tal/tal.c:482
0x5557dfc609d3 tal_free
	ccan/ccan/tal/tal.c:477
0x5557dfc609d3 towire_wally_psbt
	bitcoin/psbt.c:743
0x5557dfbc5dfc towire_dualopend_got_offer_reply
	openingd/dualopend_wiregen.c:358
0x5557dfbc5dfc openchannel2_hook_cb
	lightningd/dual_open_control.c:671
0x5557dfc22f4f plugin_hook_callback
	lightningd/plugin_hook.c:210
0x5557dfc1dfbe plugin_response_handle
	lightningd/plugin.c:591
0x5557dfc1dfbe plugin_read_json_one
	lightningd/plugin.c:702
0x5557dfc1dfbe plugin_read_json
	lightningd/plugin.c:747
0x5557dfc71756 next_plan
	ccan/ccan/io/io.c:59
0x5557dfc775d5 io_ready
	ccan/ccan/io/io.c:417
0x5557dfc775d5 io_loop
	ccan/ccan/io/poll.c:453
0x5557dfbdb1ce io_loop
	ccan/ccan/io/poll.c:380
0x5557dfbdb1ce io_loop_with_timers
	lightningd/io_loop_with_timers.c:22
0x5557dfbb37d1 main
	lightningd/lightningd.c:1195
0x7fe93ef3fd8f __libc_start_call_main
	../sysdeps/nptl/libc_start_call_main.h:58
0x7fe93ef3fe3f __libc_start_main_impl
	../csu/libc-start.c:392
0x5557dfbb6e84 ???
	???:0
0xffffffffffffffff ???
	???:0
```

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

This broke with COPTFLAGS="-flto -O3", and so I took a look (it
complains more than normal because main isn't there).  We should never
be running update-mocks except on programs expected to compile: in
this case, that's tools/test/run-test-wire.c.

Remove the code which tries to run this, which also means
non-developers won't be running update-mocks!

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

…constraints.

Get stricter with recognizing real column defs.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We can also remove the listpeers closer hack, which was removed from
the schema already.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

You often don't care about the reply, so this is quite convenient.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We have to scatter this everywhere in our schemas, as there's no way
to make it the default :(

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

valgrind noticed that this was uninitialized when I tried a complex
migration.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Because it used internal routines, it didn't pass operations through the
db hook!  So make it use the generic routines, with the twist that they
are not translated.

And when we use this in a migration hook, we're actually in a
transaction.

This, in turn, introduces an issue: we need to be outside a transaction
to "PRAGMA foreign_keys = OFF", but completing the transaction when
there is a db hook actually enters the io loop, freeing the tmpctx!

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Be more graceful in shutting down: this should fix the issue where
bookkeeper gets upset that its commands are rejected during shutdown,
and generally make things more graceful.

1. Stop any new RPC connections.
2. Stop any per-peer daemons (channeld, etc).
3. Shut down plugins.
4. Stop all existing RPC connections.
5. Stop global daemons.
6. Free up peer, chanen HTLC datastructures.
7. Close database.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Changed: Plugins: RPC operations are now still available during shutdown.