Send payments to blinded routes

[t-bast]

Since blinded routes have to be used from start to end and are somewhat similar to Bolt 11 routing hints, we model them as an abstract single hop during path-finding. This makes it trivial to reuse existing algorithms without any modifications.

We then add support for paying blinded routes. We introduce a new type of recipient for those payments, that uses blinded hops and creates onion payloads accordingly. There is a subtlety in the case where we're the introduction of the blinded route: when that happens we need to decrypt the first payload to figure out where to send the payment.

When we receive a failure from a blinded route, we simply ignore it in retries: we don't know what caused the issue so we assume it's permanent, which makes sense in most cases since we cannot change the relaying parameters (fees and expiry delta are chosen by the recipient).

By building on top of #2480, the change is quite easy to review and doesn't impact the payment state machines much. I've added a lot of e2e tests #2490, where the error handling has (finally) been correctly implemented.

This is an alternative to #2467: what I didn't like in that PR were:

• the new Partial payment onion class, which is a clear layering violation and requires reviewers to look at the whole flow of code to verify that this object contains the right data (it's quite foot-gunny because a tlv stream isn't strongly typed, it may records you don't expect...)
• the complexity of PaymentPacket.scala, which became very hard to read
• the use of a ShortChannelId in blinded hops, which was hiding subtle issues

I tried to build on top of #2467 at first, but when I realized that it was important to add the new abstraction proposed in #2480, it was impossible to do. I did pick a lot of the code from that PR though, most of the actual business logic is the same, it's just the abstractions that are different.

[thomash-acinq]

Can you add some tests to check that it actually works?

[t-bast]

Can you add some tests to check that it actually works?

I added more tests in 55f6960, let me know if there are specific test cases that would be useful that I missed. I didn't add yet an e2e test because it would require the code from #2471

I also had to add utility test functions to generate blinded routes / bolt12 invoices, those should be cleaned up when we have #2471

[thomash-acinq]

You should add the tests too now that #2499 has been merged, especially the zero-hop test which would not pass currently.

[thomash-acinq]

Looks all good.