Add new database provider for Google Cloud Spanner using the official
google-cloud-spanner library's PEP-249 DB-API interface.

Features:
- Connection via Application Default Credentials or service account JSON
- Spanner emulator support for local development
- Schema introspection via INFORMATION_SCHEMA
- Primary key detection via INDEX_COLUMNS
- Index listing and definition retrieval
- GoogleSQL dialect with backtick identifier quoting

Spanner-specific limitations handled:
- Single database per connection (no multi-database support)
- No stored procedures, triggers, or sequences
- No SSH tunneling (cloud-native service)
- INFORMATION_SCHEMA queries require read-only mode (autocommit=True)
- gRPC DNS resolver set to native for network compatibility

Note: This implementation only supports GoogleSQL dialect. Spanner databases
using PostgreSQL dialect are not supported (would require different identifier
quoting and potentially different query syntax).

Spanner databases can use either GoogleSQL or PostgreSQL dialect. This
commit adds:

- Dialect detection on connect via INFORMATION_SCHEMA.DATABASE_OPTIONS
- Dialect stored on connection object (_sqlit_spanner_dialect)
- Dialect-aware identifier quoting:
  - GoogleSQL: `identifier` (backticks)
  - PostgreSQL: "identifier" (double quotes)
- Helper methods for connection-aware quoting

Limitations:
- quote_identifier() and build_select_query() always use GoogleSQL syntax
  because they don't have access to the connection object (base class design)
- Connection-aware alternatives are provided:
  - _quote_identifier_for_conn(conn, name)
  - build_select_query_for_conn(conn, table, limit, ...)

Note: PostgreSQL dialect support is experimental and untested. The dialect
detection can be verified with GoogleSQL databases to ensure the mechanism
works correctly.