fix: update default VLM model from gemini-2.0-flash to gemini-2.5-flash

[haruyuki-oda]

gemini-2.0-flash has been discontinued by Google and returns 404 NOT_FOUND for new users. Update all default references to gemini-2.5-flash so that paperbanana generate works out of the box.

fix: update default VLM model from gemini-2.0-flash to gemini-2.5-flash

Summary

• gemini-2.0-flash has been discontinued by Google and returns 404 NOT_FOUND for new users
• Updated all default VLM model references to gemini-2.5-flash so that paperbanana generate works out of the box

Changes

File	What changed
paperbanana/core/config.py	Default in VLMConfig.model and Settings.vlm_model
paperbanana/providers/vlm/gemini.py	Default argument in GeminiVLM.__init__
configs/config.yaml	Default config
configs/provider/vlm/gemini.yaml	Gemini provider config
examples/generate_diagram.py	Example script
tests/test_providers/test_registry.py	Test assertions

Reproduction

paperbanana generate --input method.txt --caption 'Overview of our framework'
# Before: RetryError → 404 NOT_FOUND "models/gemini-2.0-flash is no longer available to new users"
# After:  Completes successfully (3 iterations, ~212s)

Test plan

• paperbanana generate --input method.txt --caption 'Overview of our framework' completes with 3 iterations
• pytest tests/ passes (53/54 pass; 1 pre-existing failure unrelated to this change)
• Verify with --vlm-model CLI override still works for users who explicitly specify a model

[haruyuki-oda]

I found that the following PR may be duplicated with me. please feel free to check and close the PR.
#45 (comment)

[dippatel1994]

#45 seems to be solving the same issue. I asked the author to close the PR, and then I will merge this PR. Thanks for your contribution.