Use correct cost comparison when evaluating candidate channels

[araspitzu]

This PR fixes bug ACINQ/eclair-mobile#207 which was caused by an incorrect channel cost evaluation against its htlcMaximumMsat. The fix is to consider the correct channel cost.

[codecov-io]

Codecov Report

Merging #1090 into master will increase coverage by 0.14%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1090      +/-   ##
==========================================
+ Coverage   82.64%   82.79%   +0.14%     
==========================================
  Files         101      101              
  Lines        7652     7653       +1     
  Branches      312      312              
==========================================
+ Hits         6324     6336      +12     
+ Misses       1328     1317      -11

Impacted Files	Coverage Δ
.../src/main/scala/fr/acinq/eclair/router/Graph.scala	97.22% <100%> (+0.01%)	⬆️
...nq/eclair/blockchain/electrum/ElectrumClient.scala	73.23% <0%> (-0.38%)	⬇️
...-core/src/main/scala/fr/acinq/eclair/io/Peer.scala	74.92% <0%> (-0.31%)	⬇️
...c/main/scala/fr/acinq/eclair/channel/Channel.scala	84.47% <0%> (+0.46%)	⬆️
...src/main/scala/fr/acinq/eclair/router/Router.scala	86.76% <0%> (+0.94%)	⬆️
...clair/blockchain/electrum/ElectrumClientPool.scala	78.49% <0%> (+4.3%)	⬆️

[sstone]

If amountMsat is not used here then it's not used at all in dijkstraShortestPath and should be removed from the parameter list. We then need to make sure that

eclair/eclair-core/src/main/scala/fr/acinq/eclair/router/Graph.scala

Line 127 in 352e4d0

val spurPath = dijkstraShortestPath(graph, spurEdge.desc.a, targetNode, amountMsat, ignoredEdges ++ edgesToIgnore.toSet ++ returningEdges.toSet, extraEdges, rootPathWeight, boundaries, currentBlockHeight, wr)

is correct

[araspitzu]

With this change dijkstraShortestPath will use only the cost found in the initialWeight parameter which is RichWeight(amountMsat, 0, 0, 0) for the initial step and when calculating the 'spurPath' for an alternative path this is the result of pathWeight so it's the cost to go from the starting node to this intermediate node (from here the algorithm will search for a different path). I will remove amountMsat from the parameter list.

[araspitzu]

f62f472 removes the unused parameter and also extracts the initialWeight parameter in the first invocation of dijkstraShortestPath, for more readability.