feat(store): add free-form tags for multi-dimensional observation filtering

[GravityZenAI]

Feature Description

The current type field is a fixed enum (bugfix, decision, architecture, etc.). In practice, observations often span multiple concerns — a bugfix that's also about auth and supabase and is critical. There's no way to express this multi-dimensionality today.

Proposal

Add an optional tags field to observations — free-form, agent-created labels for flexible categorization.

Schema change:

ALTER TABLE observations ADD COLUMN tags TEXT DEFAULT '';
-- Stored as comma-separated: "auth,supabase,critical"
-- FTS5 would index them naturally

MCP tool changes:

• mem_save(tags: "auth,critical") — attach tags when saving
• mem_search(tags: "auth,critical") — filter by tag intersection
• mem_search(tags: "auth", query: "middleware") — combine with existing text search

CLI/TUI:

• engram search --tags auth,critical
• TUI observation view shows tags as colored badges

Why it matters

Multi-dimensional filtering. An agent could ask "show me all critical auth-related observations" — impossible today without keyword-guessing in mem_search. Tags would also enable:

• Dashboard views in TUI grouped by tag
• Tag clouds showing what topics dominate a project's memory
• Cross-project tag consistency — agents could converge on shared tag vocabularies

Design principles

• Opt-in: no tags required, existing observations unaffected
• Free-form: no predefined tag list, agents create as needed
• FTS5-compatible: comma-separated text is naturally searchable
• Lightweight: single TEXT column, no join tables

Related

• Complements the Salience Scoring concept in Proposal: Five-Layer Memory Enhancement (Semantic Search, Salience, Graph, Consolidation, Priming) #168 (tags + salience = powerful filtering)
• Could integrate with topic_key for automatic tag inference

Affected Area

MCP tools, Store (schema), CLI, TUI