Implement "Simple Taproot Channels" BOLT proposal

[sstone]

This PR implements lightning/bolts#995 which introduces a new channel format where funding transactions send to an aggregated musig2 public key instead of a 2-of-2 multisig address:

• funding and closing transactions become cheaper (by about 15%)
• on-chain footprint becomes more private: funding and closing transactions are impossible to distinguish from other p2tr transactions

Basic interop testing (opening/closing channels, sending/receiving payments) pass with lnd v0.18

The trickiest part of this PR is the update to the splicing protocol: splicing means that there can be multiple commitment transactions that are active at the same time, and signatures must be exchanged for all of them. With the new musig2-based funding transaction scheme, this means generating and exchanging musig2 nonces in advance, which requires modifications to the interactive tx and splicing protocols that are not (yet) covered by the BOLT proposal.

This is still a WIP, we also need to add support for simple taproot channels to https://github.com/ACINQ/lightning-kmp.

Based on #2896

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 95.17974% with 59 lines in your changes missing coverage. Please review.

Project coverage is 86.38%. Comparing base (40f13f4) to head (51ec562).
Report is 6 commits behind head on master.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2868      +/-   ##
==========================================
+ Coverage   85.92%   86.38%   +0.46%     
==========================================
  Files         219      220       +1     
  Lines       18488    19477     +989     
  Branches      794      814      +20     
==========================================
+ Hits        15886    16826     +940     
- Misses       2602     2651      +49     

Files	Coverage Δ
...core/src/main/scala/fr/acinq/eclair/Features.scala	100.00% <100.00%> (ø)
...r/acinq/eclair/blockchain/fee/OnChainFeeConf.scala	95.65% <100.00%> (ø)
...in/scala/fr/acinq/eclair/channel/ChannelData.scala	100.00% <ø> (ø)
...cala/fr/acinq/eclair/channel/ChannelFeatures.scala	100.00% <100.00%> (ø)
...inq/eclair/channel/fsm/ChannelOpenDualFunded.scala	87.52% <100.00%> (+0.52%)	⬆️
...inq/eclair/channel/fsm/CommonFundingHandlers.scala	92.00% <100.00%> (+0.88%)	⬆️
...q/eclair/channel/publish/ReplaceableTxFunder.scala	84.81% <100.00%> (-0.53%)	⬇️
...q/eclair/crypto/keymanager/ChannelKeyManager.scala	88.88% <100.00%> (+1.38%)	⬆️
...air/crypto/keymanager/LocalChannelKeyManager.scala	100.00% <100.00%> (ø)
...air/crypto/keymanager/LocalOnChainKeyManager.scala	87.50% <100.00%> (ø)
... and 21 more

... and 5 files with indirect coverage changes

[remyers]

This looks good so far. I focused this preliminary review on splicing to identifiy any potential nonce reuse issues early.

I found one potential issue if we allow tx_abort after we send commit_sig.

I also think RBF of splices will need to be done carefully to avoid nonce resuse. I think we will need to include the RBF attempt count when generating nonces. I'll review again once you've rebased on the latest splice branch that includes splice RBF.