feat(py/plugins/google-genai): finish_message for image models and effective config echo

[cabljac]

Parallel slice of the #5806 split (based on #5854, independent of the Interactions chain): normalizes finish reasons (NO_IMAGE, IMAGE_OTHER), synthesizes finish_message when image models return no media so empty results are not silent successes, propagates finish_message onto candidates, echoes the normalized effective config on response.request, and drops the retired gemini-2.5-flash-image-preview registration.

Part of the stack splitting #5806; recombined it reproduces that PR byte-for-byte. Co-authored with @huangjeff5.

Breaking changes

• gemini-2.5-flash-image-preview is no longer registered as a model (the GEMINI_2_5_FLASH_IMAGE_PREVIEW enum members remain, but the model name stops resolving). Use gemini-2.5-flash-image.

[gemini-code-assist]

Code Review

This pull request removes the deprecated gemini-2.5-flash-image-preview model, introduces a mechanism to track and return the effective configuration (normalized to snake_case and including injected modalities) on the response's request copy, and improves handling of empty or blocked image generation responses by mapping new finish reasons (NO_IMAGE, IMAGE_OTHER) and providing descriptive finish messages. The reviewer suggests fully removing the unused GEMINI_2_5_FLASH_IMAGE_PREVIEW enum members and constants, which were left behind after removing the model from docstrings and registration.

[gemini-code-assist]

While the docstrings for VertexAIGeminiVersion and GoogleAIGeminiVersion have been updated to remove gemini-2.5-flash-image-preview, the actual enum members GEMINI_2_5_FLASH_IMAGE_PREVIEW are still defined in both enum classes, and the constant GEMINI_2_5_FLASH_IMAGE_PREVIEW remains defined in the file. To fully retire this model version, please remove these unused enum members and constants.