…ort for migrations schema and table

Move PostgreSQL-specific custom schema and table name functionality from
the Postgres implementation to the core configuration level, making it
available for all databases while respecting their schema support.

- Add database-specific configuration via `[migrate.drivers.postgres]`
- Support environment variables SQLX_MIGRATIONS_TABLE and SQLX_MIGRATIONS_SCHEMA
- Add qualified_table_name() method that returns schema-qualified names for PostgreSQL
- Update CLI to use backend_name() to detect database type
- Remove hardcoded environment variables from Postgres implementation
- Maintain backward compatibility with existing environment variables

This provides a cleaner, more maintainable approach that works across
all databases while properly handling PostgreSQL's schema support.

When loading config from a TOML file, environment variables were not being
read because the Default trait is only called for missing fields during
deserialization. This fix ensures that SQLX_MIGRATIONS_TABLE and
SQLX_MIGRATIONS_SCHEMA are checked at runtime in the qualified_table_name()
and postgres_schema() methods.

This ensures environment variables work correctly whether using
Config::default() directly or loading from a sqlx.toml file.

The migrate\! macro was expecting table_name() to return &str but it now
returns String (to support schema-qualified names). This fix converts
the String to &str before passing it to Cow::Borrowed.

This allows the migrate\! macro to properly embed schema-qualified table
names (like "my_schema._sqlx_migrations") at compile time when using
sqlx.toml configuration.

…ation

- Remove redundant qualified_table_name() method as table_name() already handles schema qualification
- Fix test compilation error by removing undefined MIGRATION_TABLE reference
- Extract duplicated schema creation logic into ensure_migration_schemas() helper
- Remove commented dead code and unrelated debug logging
- Update tests to reflect the simplified API

The table_name() method now returns the appropriate format for each database:
- PostgreSQL: schema.table (e.g., "public._sqlx_migrations")
- Other databases: table only (e.g., "_sqlx_migrations")

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

- Accept upstream connect(config, opts) signature for crate::connect()
- Preserve custom migration table_name functionality
- Fix PostgreSQL schema handling with AnyConnection backend_name()
- Update test table_name() call to use reference
- All tests passing and functionality verified