feat: add structured database seeding with YAML/JSON schema and tracking table

[mikkeldamsgaard]

Summary

Replaces the previous shell-out seed command with a declarative, structured approach that reads seed plans from YAML/JSON spec files.

Features

• Seed tracking table (initium_seed by default) for idempotent application
• Database drivers: PostgreSQL, MySQL, and SQLite
• Auto-generated IDs and cross-table references via _ref / @ref: syntax
• Environment variable substitution via $env:VAR_NAME
• Unique key detection to prevent duplicate row insertion
• Reset mode (--reset) to delete and re-apply seeds
• Transaction safety with rollback on failure
• Ordered seed sets and tables via order field

Changes

• New src/seed/ module: schema parsing, database abstraction, seed executor
• New docs/seeding.md: full schema reference, K8s usage (env vars + volume mounts), failure modes
• New example seed specs in examples/seed/
• Updated src/main.rs: new --spec and --reset CLI flags for seed subcommand
• Updated README.md, docs/usage.md, CHANGELOG.md

How to verify

cargo fmt -- --check
cargo test
cargo build

All 65 tests pass including 30+ new seed-specific tests covering schema parsing, database operations, executor logic, references, idempotency, reset mode, and edge cases.

Closes #12