feat(mcp): add RediSearch skill prompts

[joshrotenberg]

Summary

• Add three SKILL.md files under crates/redisctl-mcp/skills/ for RediSearch optimization workflows
• index-advisor: analyzes dataset shape (types, cardinality, value distributions) and recommends optimal FT.CREATE schema with aliases, weights, and SORTABLE flags
• query-tuning: uses FT.EXPLAIN and FT.PROFILE to analyze query execution plans, identify bottlenecks, and suggest optimizations with before/after comparisons
• index-ab-test: creates multiple index variants, benchmarks them with representative queries, and produces a comparison matrix to find the best configuration

These follow the Agent Skills format (YAML frontmatter + markdown body) and will be registered as MCP prompts via DynamicPromptRegistry once tower-mcp 0.8.2+ is integrated.

Tested

Both index-advisor and query-tuning were tested end-to-end against a live Redis Stack instance with 8 JSON product documents:

• Index advisor correctly inferred field types, analyzed cardinality, recommended TAG vs TEXT, and produced a working FT.CREATE with aliases
• Query tuning correctly used FT.EXPLAIN/FT.PROFILE to compare global vs field-specific queries with a metrics comparison table

Test plan

• Skills follow Agent Skills SKILL.md format
• index-advisor tested end-to-end against live data
• query-tuning tested end-to-end against live data
• index-ab-test to be validated when integrated with dynamic prompts

Ref #838