More BOLT updates, to latest.

[rustyrussell]

Includes the 12-block delay, removing non-htlc_maximum_msat updates, and setting the dont_forward bit!

Currently breaks lnprototest as expected, I'll have to fix that!

Closes: #5548
Closes #5562

[endothermicdev]

I was reading this as "Ignore any message of unknown type." Is there any other validation a reader should be expected to perform? If so, we should be explicit.

[rustyrussell]

No, that's it, so I made it more explicit.

[endothermicdev]

This was great - I learned a few things. I like the gossip_store versioning and new documentation.
ACK 6e4e104

[endothermicdev]

I should note this predictably upsets lnprototest during test_bolt7-01-channel_announcement-success.py. Will need to bring the test up to speed with a 12 block delay after closing tx.

[cdecker]

Some things that made CI upset are fixed on master so a rebase will address at least the missing mako issue.

[cdecker]

Seems there is a mismatch between bolts and string in lnprototest:

        test += [
            Block(blockheight=103, number=3, txs=[funding_tx()]),
>           ExpectMsg(
                "channel_ready",
                channel_id=channel_id_v2(local_keyset),
                second_per_commitment_point=remote_per_commitment_point(1),
            ),
            # Ignore unknown odd messages
            TryAll([], RawMsg(bytes.fromhex("270F"))),
        ]

tests/test_bolt2-20-open_channel_accepter.py:1951: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = ExpectMsg:test_bolt2-20-open_channel_accepter.py:1951
msgtypename = 'channel_ready'
if_match = <function ExpectMsg._default_if_match at 0x7f5e32b7b9a0>
ignore = None, connprivkey = None
kwargs = {'channel_id': <function channel_id_v2.<locals>._channel_id_v2 at 0x7f5e302e5510>, 'second_per_commitment_point': functools.partial(<function remote_per_commitment_point.<locals>._remote_per_commitment_point at 0x7f5e302e55a0>, 1)}

    def __init__(
        self,
        msgtypename: str,
        if_match: Callable[["ExpectMsg", Message, "Runner"], None] = _default_if_match,
        ignore: Optional[Callable[[Message], Optional[List[Message]]]] = None,
        connprivkey: Optional[str] = None,
        **kwargs: Union[str, Resolvable],
    ):
        super().__init__(connprivkey)
        self.msgtype = namespace().get_msgtype(msgtypename)
        if not self.msgtype:
>           raise SpecFileError(self, "Unknown msgtype {}".format(msgtypename))
E           lnprototest.errors.SpecFileError: (ExpectMsg:test_bolt2-20-open_channel_accepter.py:1951, 'Unknown msgtype channel_ready')

[rustyrussell]

OK, so this now depends on new lnprototest, which requires #5621...

[rustyrussell]

OK, I folded the #5621 into this (at a point before we add the 12 block delay), so this PR will actually pass.

[rustyrussell]

Trivial rebase for minor test conflict.

[cdecker]

Seems some of the blockheights are off by 12 blocks, so this seems to be either the wrong commit or we need to adjust heights in tests.

I'll be ignoring failures in the lnprototest config in CI until this PR is merged.

[rustyrussell]

Seems some of the blockheights are off by 12 blocks, so this seems to be either the wrong commit or we need to adjust heights in tests.

I'll be ignoring failures in the lnprototest config in CI until this PR is merged.

Wtf... Ok, I'll try to get to my laptop today and figure out what I did.

I meant to only add the lnprototest changes :(

[rustyrussell]

Ah, it really was a conflict! That multichan test did need fixing, done now.