This moves code that is loop invariant outside of the loop and thus will cause us to only evaluate the inequality and fee lookups once instead of every time we get to the backtrack check.
I would expect this to slightly improve the performance of the backtrack evaluation in the branch and bound algorithm.
Move the evaluation of whether the feerate is higher than the
long_term_feerate_estimate out of the loop, since it is constant across
the coin selection attempt.
This should slightly improve the performance of the backtrack evaluation
in the branch and bound algorithm.
co-authored-by: Yancy Ribbens <email@yancy.lol>
Nevermind, should be an improvement to #25932