Subdifferential discrepancy calculation scaling fix for sliding PDPS

Store subdifferential discrepancy in log

Small optimisation

attempt to reduce trait bounds

Relax step length bounds

Extra weight optim was probably disabled

Generalise algorithms to arbitrary domains

target_bounds not needed

Do not prune new insertions on first round

Added greedy insertion option, but does and actually really should not too much

(0) -10 tip

mercurial