[Splicing] Clone for ChannelContext

[optout21]

Cloning of a channel -- more precisely of ChannelContext -- is needed for Splicing, this is a preparation (may be needed for dual funding RBF as well)

Simple #[derive(Clone)) is not sufficient, as the channel signer struct cannot be cloned, ultimately due to AtomicUSize and Secp256k1 (in KeysManager). So instead, a field-by-field cloning is done, with the a few exceptions.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.00%. Comparing base (ba3d4ff) to head (1cdce16).
Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3332      +/-   ##
==========================================
+ Coverage   89.25%   90.00%   +0.74%     
==========================================
  Files         130      130              
  Lines      106959   111140    +4181     
  Branches   106959   111140    +4181     
==========================================
+ Hits        95464   100029    +4565     
+ Misses       8706     8343     -363     
+ Partials     2789     2768      -21     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jkczyz]

Discussed briefly offline. This may not be needed if we split ChannelContext to avoid duplicating some fields.

[optout21]

This change request assumes that the whole ChannelContext struct is duplicated, which may not be the best solution. I put this on hold for now (back to draft, description updated).

[jkczyz]

Not sure if cloning is the best solution. But, regardless, since the channel signer is a Deref can't we just add a Clone trait bound?

[optout21]

There are 4 Mutex fields, used for testing/fuzzing: I placed them under refcounting (`Arc), so they can be cloned.

The only non-cloneable field is the signer. The signer could also be placed under Arc, but I find that unjustified complexity and inefficiency for this infrequent use case.