Improve blinded route data structure

[thomash-acinq]

Change the blinded route structure to better suit our needs:

• We need the blinded node ids of all nodes, including the introduction node.
• We only need the blinding ephemeral key for the first node, not the other ones. And in the case where we didn't build the blinded route ourselves we can't have access to them anyway. By only storing the first ephemeral key we can use the same data structure to represent blinded routes that we receive from our peers.

[codecov-commenter]

Codecov Report

Merging #2025 (8328a09) into master (bdef833) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2025      +/-   ##
==========================================
+ Coverage   87.70%   87.74%   +0.04%     
==========================================
  Files         161      161              
  Lines       12562    12560       -2     
  Branches      534      524      -10     
==========================================
+ Hits        11017    11021       +4     
+ Misses       1545     1539       -6     

Impacted Files	Coverage Δ
...src/main/scala/fr/acinq/eclair/crypto/Sphinx.scala	100.00% <100.00%> (ø)
...ala/fr/acinq/eclair/balance/ChannelsListener.scala	93.10% <0.00%> (-3.45%)	⬇️
...c/main/scala/fr/acinq/eclair/channel/Channel.scala	88.83% <0.00%> (-0.08%)	⬇️
...clair/channel/publish/ReplaceableTxPublisher.scala	87.35% <0.00%> (+1.14%)	⬆️
...main/scala/fr/acinq/eclair/router/Validation.scala	92.30% <0.00%> (+1.53%)	⬆️
...cala/fr/acinq/eclair/payment/relay/NodeRelay.scala	97.56% <0.00%> (+1.62%)	⬆️

[t-bast]

And in the case where we didn't build the blinded route ourselves we can't have access to them anyway.

Good point!

I don't think we should throw the ephemeral keys away though, other schemes than onion messages may need it and it's useful to test them against reference test vectors (and I don't like throwing information away).

I updated my initial PR to separate the ephemeral keys from the route itself, let me know what you think.
Also it seems like you've removed the encrypted payload for the introduction node here, was that intended?

[thomash-acinq]

Also it seems like you've removed the encrypted payload for the introduction node here, was that intended?

No, it's still there: blinded.route.encryptedPayloads.head

[t-bast]

Oh that's because you put the introduction node back into the blindedNodes field. I think it's messy as it has its data split between its BlindedNode representation and the additional fields you added. I prefer explicitly singling it out, WDYT?

[thomash-acinq]

Can you give me an example usage where you would want to process the introduction node separately? For onion messages I don't.
I'm actually thinking that the val nodeIds: Seq[PublicKey] = introductionNodeId +: blindedNodes.tail.map(_.blindedPublicKey) is not useful and should be replaced by val blindedPublicKeys: Seq[PublicKey] = blindedNodes.map(_.blindedPublicKey)