Fix copy/paste bugs in duplicated conditions

[lyskov-ai]

Each condition changed here tests the same operand twice, so a sibling operand is never tested and part of the check is dead code.

Behaviour fixes

core/energy_methods/CartesianBondedEnergy.cc — eval_singleres_torsion_derivatives tests whether all four torsion atoms are ring atoms before handing the torsion off to cart_bonded_ring. It checked atids[2] twice and never atids[4]. The two sibling angle loops in the same file (over a Size3) correctly check atids[1..3]; this one is over a Size4.

core/chemical/rna/util.cc — in the SUGAR/SUGAR CIS block of get_base_pair_atoms, the third row of the 4×4 base-pair table is labelled aa1 == na_rad, which the adenine row above it already matches, so all four of its branches are unreachable. They belong to guanine:

• the bodies push " H22", an atom that only RGU has — RAD has " H2 " (see RAD_n.params / RGU_n.params);
• na_rgu is the only aa1 value missing from the chain, which otherwise covers rad, rcy, ura against all four aa2 values.

protocols/minimization_packing/DisulfideOptimizationMover.cc — the check documented as "Confirm that both are cysteine-type residues in a disulfide bond to each other" tested cys_pos[1] twice, so cys_pos[2] was never validated. The error messages below it already reference both residues.

protocols/environment/DofUnlock.cc — the environment-id half of the popped-passport check compared pass_out to itself, so an environment-id mismatch was never caught. The error message below it reports pass_->env_id() as the expected value.

protocols/pose_metric_calculators/RotamerRecovery.cc — runtime_assert( pose_rots.size() == pose_rots.size() ) is always true; the neighbouring assert compares against the native, and this one should too.

protocols/floppy_tail/FloppyTailMover.cc — flexible_chain appeared twice in the list of options incompatible with an explicit -in:file:movemap, so flexible_start_resnum was never rejected even though the movemap overrides it. The user-facing message had the same duplication and is corrected to match.

Behaviour-preserving removals

These duplicates are simply dead and removing them changes nothing:

• protocols/cryst/wallpaper.cc — "C211" listed twice among the tetragonal groups (the file defines exactly seven distinct groups for that setting).
• protocols/antibody/design/CDRSeqDesignOptions.cc — "DISALLOWED" listed twice.
• protocols/simple_filters/SecretionPredictionFilter.cc and protocols/protein_interface_design/movers/SecretionOptimizationMover.cc — the dG_ins threshold comparison repeated in the same && chain, left over from extending the local-minimum test from ±1 to ±2.
• protocols/sic_dock/xyzStripeHashPoseWithMeta.hh — init_with_pose tested PoseCoordPickMode_BB twice, making the second branch unreachable. Its body collects CA/C/CB, which matches no value of PoseCoordPickMode (N_CA_C and N_CA_C_CB both include N), so the intent behind it is not recoverable and the dead branch is removed. The surviving branch collects N/CA/C/O/CB, matching both the natom counting loop directly above it and the canonical case PoseCoordPickMode_BB in core/pose/xyzStripeHashPose.cc.

Note on diff size

The diff is small relative to the usual bundling guidance because every fix is a one-line correction and this category is exhausted: the analysis pass was run over utility/, basic/, core/, protocols/ and numeric/, and utility/, basic/ and numeric/ produced no true positives. The remaining flagged sites are either the x != x NaN idiom or genuine "different conditions, same action" branches.