docs(readme): shorten 2026-05-16 image-handling-parity news item

[ericleepi314]

Summary

The 2026-05-16 image-handling-parity news entry I landed in #157 was a paragraph long; every other News bullet is one or two sentences. Trimming it to match the house style: four facts joined by semicolons, no inline rationale.

Before

Read tool ported the TS image pipeline (magic-byte format sniff, bounded byte read, resize/compress to the 3.75 MB / 1568 px envelope, base64 size validation); `@image.png` @-mentions now inline as real `ImageBlock` content (the prior text-mode read shipped mojibake the model latched onto); image `tool_result` list shape preserved end-to-end through `_dispatch_single_tool`; OpenAI-compat / GLM / Minimax / DeepSeek / OpenRouter now translate Anthropic `image` and `document` blocks into `image_url` / `file` shapes (with `tool_use_id` correlation across the tool→user split); `validate_images_for_api` promoted into `BaseProvider._prepare_messages` so every provider validates client-side; BOM-aware text decoding (`utf-16` / `utf-32` / `utf-8-sig`) handles Windows-emitted Unicode files without mojibake.

After

Read tool TS image pipeline (sniff, resize/compress, base64 cap); `@image.png` mentions inline as real `ImageBlock`s instead of mojibake; cross-provider Anthropic `image`/`document` → OpenAI `image_url`/`file` translation; pre-API base64 size validation in `BaseProvider`.

The detailed write-up still lives in the PR descriptions for #149 / #154 / #155 / #156 for anyone who wants it.

Test plan

• Diff is README-only (+1/-1).

🤖 Generated with Claude Code