Loothing

This release is a performance-focused hardening pass for the live loot-council flow. It keeps vote and response traffic batched instead of multiplying messages per viewer, while tightening the edge cases found in the latest raid reports.

Fixed

• Council votes no longer create per-viewer update traffic. Vote updates stay as one ML-originated batched update per changed candidate, avoiding the excessive addon-channel load from private per-recipient vote messages.
• Large response bursts no longer disappear. When many raiders answer at once, Loothing now splits candidate-update batches to fit the receive-side batch limit instead of sending an oversized batch that clients reject.
• The master looter's own multi-item responses are processed once. If the ML responds to several items at once, Loothing now handles the local responses without also self-sending and double-counting them.
• ML Observer Mode now applies consistently to council voting. When the master looter is configured as an observer, they are excluded from vote prompts, vote polls, vote resends, and accepted council vote commits.
• Duplicate copies of the same item are preserved more reliably. Bag scans now distinguish the second physical copy from the roll-won refresh path, so identical drops do not collapse into one row.
• Old trade-window inventory no longer looks like new boss loot. Pre-pull bag snapshots count existing tradeable items normally, then the post-kill scan reports only item IDs whose bag count actually increased.
• Bag scans do less work each tick. The scanner now counts item IDs first and only reads links/tooltips for item IDs that actually changed, reducing repeated per-slot allocation during the 60-second scan window.
• Hidden vote and response views are more consistent. Results, item rows, council-table progress, and observer views now avoid exposing hidden vote counts or response ordering through secondary UI text or sorting.