o4-mini: reasoning.summary parameter blocked by upstream OpenAI organization verification

### Summary

When calling `o4-mini` via the Responses API on a Foundry project endpoint with `reasoning={"summary": "auto"}` (or `"concise"` / `"detailed"`), the request fails with HTTP 400 and an OpenAI-style error message asking the *organization* to be verified:

```
BadRequestError: Error code: 400 - {'error': {'message': 'Your organization must
be verified to generate reasoning summaries. Please go to:
https://platform.openai.com/settings/organization/general and click on Verify
Organization. If you just verified, it can take up to 15 minutes for access to
propagate.', 'type': 'invalid_request_error', 'param': 'reasoning.summary',
'code': 'unsupported_value'}}
```

This error comes from upstream OpenAI infrastructure — but from a Foundry customer's perspective it's surfaced through their Foundry project endpoint with no clear remediation path, since they don't have access to a `platform.openai.com` organization to verify.

### Reproduction

```python
from azure.identity import DefaultAzureCredential, get_bearer_token_provider
from openai import OpenAI

endpoint = ""
client = OpenAI(
base_url=endpoint.rstrip("/") + "/openai/v1",
api_key=get_bearer_token_provider(DefaultAzureCredential(), "https://ai.azure.com/.default"),
)

# Works — emits reasoning item with reasoning_tokens > 0, but summary[].text is empty
r = client.responses.create(
model="o4-mini",
input="What's heavier: a pound of feathers or a pound of bricks? Think step by step.",
max_output_tokens=2000,
)
print(r.usage.output_tokens_details.reasoning_tokens) # >0
for item in r.output:
if item.type == "reasoning":
print([s.text for s in item.summary]) # ['']

# Fails — HTTP 400 'organization must be verified'
r = client.responses.create(
model="o4-mini",
input="...",
max_output_tokens=2000,
reasoning={"summary": "auto"},
)
```

### Observed behavior

- Default request (no `reasoning` param): `o4-mini` emits a structured `reasoning` output item with `reasoning_tokens > 0` reliably, **but** `summary[].text` is empty
- Adding `reasoning={"summary": "auto"|"concise"|"detailed"}`: HTTP 400 with the org-verification message above

### Expected behavior

- Either (a) summary text should be returned by default for verified Foundry deployments without requiring `platform.openai.com` org verification, or (b) the README/docs/examples should document the constraint and a remediation path
- The error message, when shown via Foundry, should point at a Foundry/Azure remediation flow rather than `platform.openai.com` which Foundry customers may not have access to

### Impact

- Customers using OpenAI reasoning models (`o4-mini`, presumably `o4`, `o3`, etc.) through Foundry can't surface reasoning summaries to end users, even though Foundry is billing them for the reasoning tokens
- The error message is misleading in a Foundry context — customers may have no way to "verify their organization" at `platform.openai.com`
- Templates that surface reasoning summaries (like #11) work for non-OpenAI reasoning models but silently come back with empty summaries for o-series models — confusing for demo authors

### Suggested fix

Either:
1. Treat verified Foundry resources as implicitly org-verified for reasoning summary access
2. Surface a Foundry-specific error message pointing to docs that explain the constraint
3. Document the constraint clearly in the README / known issues so customers don't burn time investigating

### Environment

- Foundry account in East US 2
- `o4-mini` deployment, version `2025-04-16`
- `openai==2.36.0`, `azure-identity==1.x`
- Project endpoint, `/openai/v1` base URL, EntraID `DefaultAzureCredential` with `https://ai.azure.com/.default` scope

### Cross-reference

PR #11 surfaces reasoning summaries across all 5 language examples. The current behavior means OpenAI reasoning models silently return empty summary text through the universal pattern, which is acceptable but worth documenting.