Fix: Allow 'tool' role in ChatMessageInput

[otchlan]

Problem

Fixes #486

The ChatMessageInput interface did not allow the tool role, even though this role was defined in ChatMessageRoleData and is necessary for proper tool calling workflows.

This prevented users from:

• Importing/exporting chat histories with tool calls
• Continuing conversations after tool execution
• Building agents that use tools locally

Solution

Changed ChatMessageInput to use ChatMessageRoleData type instead of hardcoded union type:

• Updated role field to use ChatMessageRoleData
• Updated chatMessageInputSchema to use chatMessageRoleDataSchema

Testing

Created a reproduction test that verifies Chat.from() now accepts messages with role: "tool":

const chat = Chat.from([
  { role: "user", content: "Can you search for something?" },
  { role: "assistant", content: "Sure, I'll search for that." },
  { role: "tool", content: "Search results: ..." }  // ✅ Now works!
]);

Changes

• packages/lms-client/src/ChatInput.ts: Updated interface and schema

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Tool role converted to wrong content shape

Allowing ChatMessageRoleData in ChatMessageInput means callers can now pass { role: "tool", content: "..." }, but chatMessageInputToChatMessageData still always builds text/file parts and then forcibly casts the result to ChatMessageData here. Per ChatHistoryData.ts (lines 139‑157) tool messages must contain toolCallResult parts, so a tool message created through Chat.from now produces an invalid ChatMessageData object. Any downstream validation with chatMessageDataSchema (e.g., when cloning a ChatMessage) or the LLM request path will throw or drop the tool result because the expected toolCallResult part is missing.

Useful? React with 👍 / 👎.

[otchlan]

I have read the CLA Document and I hereby sign the CLA