Python: feat(anthropic): Add response_format support for structured outputs

[sukeesh]

Motivation and Context

The Anthropic client was missing response_format handling, which is
supported by OpenAI and Azure AI clients. This adds structured output
support using Anthropic's tool use feature.

Changes:

• Add _create_response_format_tool to convert Pydantic models to tool definitions
• Force model to use the response format tool when response_format is provided
• Convert tool call arguments to TextContent for JSON parsing
• Pass response_format to ChatResponse constructor to trigger try_parse_value

This enables consistent response_format usage across all chat clients.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the Contribution Guidelines
• All unit tests pass, and I have added new tests where possible
• Is this a breaking change? If yes, add "[BREAKING]" prefix to the title of the PR.

[eavanvalkenburg]

LGTM, thanks!

[markwallace-microsoft]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/anthropic/agent_framework_anthropic
_chat_client.py	360	151	58%	51, 361, 384, 386, 401, 423–426, 435, 437, 468–472, 474, 476–477, 479, 484–485, 487, 518–519, 528, 530–531, 536, 553–554, 596, 611, 615–616, 632, 641, 643, 647–648, 691–693, 695, 708–709, 716–718, 722–724, 728–731, 742, 744, 766, 776, 798–804, 811–812, 820–821, 829–832, 839–840, 846–847, 853–854, 860, 868–870, 874, 881–882, 888–889, 895–896, 902, 910–913, 920–921, 940, 947–948, 967, 989, 991, 1000–1001, 1007, 1029–1030, 1036–1037, 1046–1056, 1063–1069, 1076–1082, 1089–1098, 1105–1108
TOTAL	17473	2712	84%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
3186	213 💤	0 ❌	0 🔥	1m 9s ⏱️

[sukeesh]

@dmytrostruk / @victordibia / @moonbox3 Can I get your review as well on this?

[sukeesh]

@dmytrostruk @eavanvalkenburg Can you please re-review? I fixed the ruff lint issue that it was complaining.

[sukeesh]

cc @moonbox3