Optimize cold start of neptune-query

[PatrykGala]

This change reduces client startup latency by eliminating initial config/discovery network calls during get_client() initialization.
Instead of fetching client_id and token URLs from server config, auth refresh metadata is derived directly from OAuth token claims.

What changed

• Simplified client bootstrap:
  • Removed get_config_and_token_urls() and TokenRefreshingURLs.
  • internal/client.py now creates the auth client directly from credentials.
• Simplified auth client constructor path:
  • Removed client_id and token_refreshing_endpoint from AuthenticatedClient/NeptuneAuthenticator.
• Moved refresh metadata to token payload:
  • OAuthToken now stores:
    • client_id from azp
    • token_endpoint from iss + /protocol/openid-connect/token
  • Refresh flow now uses self._token.token_endpoint.
• Updated generated code templates and mirrored generated files.
• Updated tests and fixtures (unit/e2e/perf) to match new auth model.

Impact

• Fewer network calls on cold start.
• Faster initial client creation.
• Refresh endpoint/client identity now come from JWT claims (iss, azp) instead of config discovery.