Embed friendly instructions in bundled GPT-5.4 and GPT-5.5 (#44930)

## What changed

Replace the selectable personality templates in the bundled `gpt-5.4` and `gpt-5.5` definitions with fixed friendly instructions. Personality selection becomes unavailable for these models in the TUI, and submitted turns omit the personality override. Remove personality overrides from Python SDK examples and the walkthrough notebook.

## Testing

Update core and TUI tests to cover fixed friendly instructions, ignored pragmatic updates, unavailable personality selection, and omitted turn overrides. Preserve coverage for selectable personalities in legacy catalogs, and add coverage for resumed legacy sessions and explicit empty instruction overrides.

GitOrigin-RevId: 869a71445584272405fcbd1325efc31c224808c8

sdk/python/examples/12_turn_params_kitchen_sink/async.py

@@ -16,7 +16,6 @@ from openai_codex import (
     AsyncCodex,
 )
 from openai_codex.types import (
-    Personality,
     ReasoningSummary,
 )
 
@@ -50,7 +49,6 @@ async def main() -> None:
         turn = await thread.turn(
             PROMPT,
             output_schema=OUTPUT_SCHEMA,
-            personality=Personality.pragmatic,
             summary=SUMMARY,
         )
         result = await turn.run()

sdk/python/examples/12_turn_params_kitchen_sink/sync.py

@@ -14,7 +14,6 @@ from openai_codex import (
     Codex,
 )
 from openai_codex.types import (
-    Personality,
     ReasoningSummary,
 )
 
@@ -44,7 +43,6 @@ with Codex(config=runtime_config()) as codex:
     turn = thread.turn(
         PROMPT,
         output_schema=OUTPUT_SCHEMA,
-        personality=Personality.pragmatic,
         summary=SUMMARY,
     )
     result = turn.run()

sdk/python/examples/13_model_select_and_turn_params/async.py

@@ -16,7 +16,6 @@ from openai_codex import (
     Sandbox,
 )
 from openai_codex.types import (
-    Personality,
     ReasoningEffort,
     ReasoningSummary,
 )
@@ -100,7 +99,6 @@ async def main() -> None:
             effort=selected_effort,
             model=selected_model.model,
             output_schema=OUTPUT_SCHEMA,
-            personality=Personality.pragmatic,
             sandbox=Sandbox.read_only,
             summary=ReasoningSummary.model_validate("concise"),
         )

sdk/python/examples/13_model_select_and_turn_params/sync.py

@@ -14,7 +14,6 @@ from openai_codex import (
     Sandbox,
 )
 from openai_codex.types import (
-    Personality,
     ReasoningEffort,
     ReasoningSummary,
 )
@@ -95,7 +94,6 @@ with Codex(config=runtime_config()) as codex:
         effort=selected_effort,
         model=selected_model.model,
         output_schema=OUTPUT_SCHEMA,
-        personality=Personality.pragmatic,
         sandbox=Sandbox.read_only,
         summary=ReasoningSummary.model_validate("concise"),
     ).run()