Fix "script ran too long" in ScanExistingItems via lookup tables

[ptarjan]

Summary

Build spellId / creatureId lookup tables before iterating collections in ScanExistingItems(), converting the inner loops from O(groups * items) per collection entry to O(1) hash lookups.

Fixes #554, #753, #878, #895

Problem

ScanExistingItems() has an O(n*m) nested loop: for each collected mount/pet/companion (~2000+), it iterates through ALL items in R.db.profile.groups to find matches by spellId or creatureId. With 2175 pets and hundreds of tracked items, this produces millions of iterations in a single frame, triggering the "script ran too long" error.

Solution

Before iterating collections, pre-build index tables from R.db.profile.groups:

• spellIdIndex[spellId] → list of item entries (for mounts/companions)
• creatureIdIndex[creatureId] → list of item entries (for pets)

The mount, companion, and pet inner loops are replaced with O(1) hash lookups via ipairs over the matching entries.

Complexity improvement:

• Before: O(collections × groups × items) per scan
• After: O(groups × items) to build index once + O(collections) to scan

Test plan

• /reload with 2000+ collected pets — no more "script ran too long"
• Rarity correctly marks owned mounts/pets as found
• Tooltip still shows collection status for tracked items