[Sad Catapella B4][Performance] Design reconciliation and duplicate checks repeatedly scan all rows #85

Baseline and classification

Sad Catapella Batch 4, commit d7a538e56d41ac1093ba79adcad397c21b3dfd32. Confirmed O(N²)-class algorithmic work, not a measured bottleneck.

Evidence

• reconcile_design() lines 401–430 loops over each current biological repeat and calls previous.row(repeat.key), which scans all design rows.
• ExperimentDesign.conflicts lines 192–204 scans group members and repeatedly calls conditions.count(c) within each group. members() performs another full scan.
• Both lenient duplicate-key check and strict duplicate-key check repeatedly run keys.count(key) inside a per-key comprehension.

Impact

Slow design loading, reconciliation and confirmation as biological-repeat count grows. This is separate from N-image histogram memory (#77) and mask-filename matching (#84 if numbered accordingly).

Fix/acceptance

Build key→row dictionary once per reconciliation; precompute group→condition counters for conflicts; use collections.Counter/a single seen-key pass for duplicates, preserving error order and strict/lenient semantics. Benchmark 100/1k/10k biological repeats with adversarial group sizes; assert exact issue messages and deterministic order. Numba not appropriate.