Add Snowflake rules: SQL development, data pipelines, and Cortex AI

[jamescha-earley]

Summary

Three focused Snowflake rule sets for the rules collection:

• snowflake-sql-development -- Naming conventions, stored procedure colon-prefix gotcha, semi-structured data patterns, MERGE upserts, performance tuning (cluster keys, search optimization, warehouse sizing), common error patterns, and security essentials.
• snowflake-data-pipelines -- Dynamic Tables vs Streams+Tasks decision framework with blocker table, DT creation and design rules, Snowpipe, pipeline debugging queries, dbt on Snowflake configs, and pipeline anti-patterns.
• snowflake-cortex-ai -- Cortex AI function reference (with deprecated name warnings), TO_FILE argument rules, AI_CLASSIFY vs AI_COMPLETE guidance, Cortex Search service creation, Cortex Agent spec structure rules, Snowpark Python session setup and vectorized UDFs, and cost estimation patterns.

Each rule uses globs **/*.sql and alwaysApply: false. README updated with a new Data Platform category section.

No existing rules were modified.

Why Snowflake?

Snowflake has significant syntax gotchas that trip up AI coding agents -- the stored procedure colon prefix rule, the TO_FILE two-argument signature, Dynamic Table constraints, and the $spec$ vs $$ delimiter distinction for agents. These rules prevent the most common mistakes.