Fix pyconfig strict validation for multimodal keys in text-only runs

[SurbhiJainUSC]

Description

Fix pyconfig strict validation for multimodal keys in text-only runs

The Pydantic configuration validation in pyconfig.py was overly strict for text-only workflows (such as RL training). When processing model YAMLs that included multimodal parameters (like vision_encoder_block or image_size_for_vit), it would crash with a ValueError because those keys are not always in the active schema.

This commit updates _prepare_for_pydantic to check the use_multimodal flag. If use_multimodal is False, the parser now gracefully ignores these specific vision keys and drops them instead of crashing.

FIXES: b/538001451

Tests

Local run

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.