feat(py): fallback to BACKGROUND_MODEL in resolve_model

[huangjeff5]

Summary

• resolve_model looks up MODEL, then BACKGROUND_MODEL. A name registered only with define_background_model is findable under the same string callers already pass.
• A foreground model still wins when both kinds share a name.
• Missing names stay None (NOT_FOUND at the veneer).
• A plugin whose resolve(MODEL) is None still yields the BACKGROUND_MODEL start action for a slashed name (plug/veo-…).

This PR is lookup only. generate() / generate_operation() boxing is #6130. Plugin resolve(MODEL, veo) already returns None on this stack, so the fallback can see Veo once that name is asked for.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Code Review

This pull request introduces robust support for ModelRef and pure model resolution helpers in the Genkit Python SDK, allowing model names, versions, and configurations to be resolved, normalized, and merged cleanly. It updates core APIs like generate, generate_stream, generate_operation, define_prompt, and define_agent to accept ModelRef and handle configuration aliases (camelCase to snake_case). Additionally, it updates reflection endpoints and registry lookups, and adds comprehensive unit tests. Feedback is provided regarding a potential misleading error in resolve_model_arg when the default model is configured as an empty string.