Fix Gemini response modalities to include TEXT and IMAGE (#125)

* fix(images): use TEXT+IMAGE responseModalities for Gemini models

The Gemini API's documented default is responseModalities: ["TEXT", "IMAGE"].
The previous hardcoded ["Image"] worked for gemini-2.5-flash-image but
fails for gemini-3-pro-image-preview (a thinking model that requires
both modalities).

This aligns with Google's API documentation and ensures compatibility
with both current and future Gemini image models.

Fixes #123

https://claude.ai/code/session_01KYduqFZTvWMNMBW9b1nLXF

* chore: bump version to 0.9.4

https://claude.ai/code/session_01KYduqFZTvWMNMBW9b1nLXF

---------

Co-authored-by: Claude <noreply@anthropic.com>

Author: Kamilbenkirane

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "celeste-ai"
-version = "0.9.3"
+version = "0.9.4"
 description = "Open source, type-safe primitives for multi-modal AI. All capabilities, all providers, one interface"
 authors = [{name = "Kamilbenkirane", email = "kamil@withceleste.ai"}]
 readme = "README.md"

src/celeste/modalities/images/providers/google/gemini.py

@@ -84,7 +84,7 @@ def _init_request(self, inputs: ImageInput) -> dict[str, Any]:
         return {
             "contents": [{"parts": parts}],
             "generationConfig": {
-                "responseModalities": ["Image"],
+                "responseModalities": ["TEXT", "IMAGE"],
                 "imageConfig": {},
             },
         }