Chat: Show provider instance name for duplicate BYOK models in model picker

[fishcharlie]

Summary

This PR updates Chat model picker labeling for grouped language model providers so duplicate models from the same vendor are distinguishable by provider instance name.

Specifically, when resolving grouped models, the displayed model detail now uses the configured provider group name (for example, Local MacBook, Remote Linux) instead of the generic vendor label (for example, Ollama).

Fixes #312025

Problem

When users configure multiple Ollama providers and those providers expose overlapping model names, the model picker showed the same generic label for each duplicate row. That made it unclear which backend was being selected.

Change

• In grouped language-model resolution, model detail is overridden with the user-configured provider group name.
• This preserves unique context per provider instance in the picker while keeping the rest of model metadata behavior unchanged.

Why this approach

• Centralized fix in grouped model resolution path.
• Works for duplicate models across multiple instances of the same vendor.
• Avoids provider-specific special handling.

Tests

Automated

• Added unit test suite: LanguageModels - Provider Group Detail Override.
• Test validates that two groups from the same vendor resolve the same model id with different detail values matching group names.

Manual

1. Configure two Ollama providers with different names.
2. Ensure both providers expose at least one duplicate model.
3. Open Chat model picker.
4. Verify duplicate model rows now show provider instance names (group names), not a generic Ollama label.

Screenshots

Before

After

Risk and impact

• Low risk: change is limited to grouped model metadata detail in picker-related model resolution.
• No expected behavior change for non-grouped providers.

[Copilot]

Pull request overview

This PR improves Chat’s model picker labeling for grouped language model providers so that duplicate models from the same vendor can be distinguished by the configured provider instance (group) name.

Changes:

• Override grouped language model metadata.detail to use the provider group name during grouped model resolution.
• Add a unit test validating that the same model ID exposed by multiple groups resolves with distinct detail values matching the group names.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
src/vs/workbench/contrib/chat/common/languageModels.ts	Overrides grouped model metadata.detail with the configured provider group name during model resolution.
src/vs/workbench/contrib/chat/test/common/languageModels.test.ts	Adds a unit test ensuring grouped instances produce distinct detail values for duplicate models.

[lramos15]

How do we feel about just not setting detail in the chat extension and then if detail isn't set we set it to Group name. That way we aren't making assumptions that extensions are setting vendor = detail or having to iterate the model list creating a clean separation of concerns by basically providing a reasonable default

[fishcharlie]

@lramos15 Should be addressed in dd70562

[fishcharlie]

@lramos15 Seems like this broke in a recent Insiders build.