Addon: [1.9.6] - TextReader allows UTF-8 characters extraction

[Xian55]

Replace ChatReader with unified TextReader that supports UTF-8 text transfer from the addon via cells 98-99. The new encoding packs 3 raw UTF-8 bytes per frame into a 24-bit data cell with metadata encoding:

• cmd(4 bits) | length(10 bits) | offset(10 bits).

TextReader handles multiple text command types through a single channel:

• Chat messages (whisper, say, yell, emote, party)
• Target names (for player targets not in CreatureDB)
• Totem names (from combat log detection)

Add TotemDetector that cross-references CombatLog.EnemySummons against CreatureDB to identify Totem creature types. Detection uses two methods for Classic client compatibility:

• SPELL_SUMMON from hostile NPCs (standard client versions)
• SPELL_AURA_APPLIED from hostile Guardians (Classic 1.14.x where NPC totems appear as COMBATLOG_OBJECT_TYPE_GUARDIAN 0x2000)

Expose "DamageTakenFromTotem" requirement in RequirementFactory for use in class config JSON files to condition-gate spells based on totem presence (e.g. interrupt channeled casts, trigger defensive cooldowns).

Lua-side changes:

• Unified TextQueue with free-list entry pooling (no allocation on push)
• UTF-8 safety check and 4-byte sequence filtering (emoji)
• PackUTF8Bytes helper for zero-allocation byte packing
• Destructure sourceName/destName from combat log args directly instead of using select() to avoid off-by-one indexing errors

C#-side changes:

• CombatLog tracks EnemySummons HashSet via new cell 110
• AddonReader falls back to TextReader.LastTargetName for player targets instead of the old 6-char ASCII cells 16-17
• ChatReader deleted, TextReader registered as IReader singleton
• Frame count 113 -> 114, addon version 1.9.5 -> 1.9.6