fix: include input_tokens in message_delta events for Anthropic compatibility

[dracpet]

Problem

Some Anthropic-compatible clients (e.g., AgentScope/QwenPaw) expect input_tokens to be present in the message_delta event's usage object. Currently, stream_from_openai_result only includes output_tokens, which causes TypeError: '>' not supported between instances of 'NoneType' and 'int' in frameworks that parse usage.input_tokens from the final delta event.

Solution

Add input_tokens to the usage dict in the message_delta event, matching the value from message_start.

Changed

# Before
"usage": {"output_tokens": usage.get("completion_tokens", 0)},

# After  
"usage": {
    "input_tokens": usage.get("prompt_tokens", 0),
    "output_tokens": usage.get("completion_tokens", 0),
},

Verification

Tested with:

• ✅ Simple streaming & non-streaming
• ✅ Tool definitions & tool results
• ✅ Multi-turn conversations (3 turns)
• ✅ Large system prompts (skill injection simulation)
• ✅ Minimal responses (edge case)
• ✅ Stop sequences
• ✅ Concurrent requests (3 simultaneous)
• ✅ AgentScope/QwenPaw integration

All 10 test scenarios pass without regression.

[Upper9527]

The compatibility fix still looks relevant, but this branch now conflicts with the substantially updated Anthropic protocol adapter. Please rebase onto current main and add a regression test asserting that the final message_delta.usage contains both input_tokens and output_tokens. Once rebased, this should be a small change to review.

[dracpet]

Rebased onto current main (3938f75) and added the regression test.

• Final message_delta now emits "usage": {**anthropic_usage} — same shape as message_start: input_tokens + output_tokens (+ cache_read_input_tokens when prompt caching is reported).
• New test test_streaming_final_message_delta_usage_includes_input_and_output_tokens asserts the final message_delta.usage carries both fields.
• pytest tests/test_anthropic_messages.py → 19 passed, ruff check clean.

Note: the full suite has 17 pre-existing failures in test_responses_native.py (identical on clean main, so unrelated to this diff); left untouched.