优化 Issue Tracker Coordinator 的行为

Two usability/reliability problems observed in the Issue Tracker agent-team (agent-teams/issue-tracker/) while using it to track and delegate an unrelated bug report.

1. Sub-agent Task/Run note has no readable title — it echoes the full goal text as the label

When the Issue Tracker orchestrator creates a durable Task/Run for a delegated sub-agent (via POST task/create), the generated Task Note and the Run's root question node both use the entire (long) goal string as their auto-derived label, e.g.:

"Investigate root cause of two GitHub Copilot model-provider bugs in Huabu: (1) auth error 'Credential store modify failed for github-copilot' during model settings setup, and (2) the device-login..."

This makes the canvas hard to scan: node titles are long, truncated, and don't distinguish one delegated Run from another at a glance.

Proposed direction: allow (or require) the calling Agent to pass a short, human-readable title (e.g. the issue name/number, such as "Issue #NN: GitHub Copilot auth + device code") separately from the full goal text, and use that title as the Task Note / Run node label instead of the raw goal. This likely touches the task/create request contract (an optional title field alongside goal) and the RFS skill/tasks guide/documentation.

2. Configuration error when the Issue Tracker profile is used as the sub-agent Profile for its own delegated Run

When starting an investigation Run with defaultRootProfileId / rootProfileId set to the Issue Tracker's own Profile ID (i.e., using issue-tracker as the sub-agent Profile instead of an external coding Profile), a configuration error was encountered. Root cause has not yet been analyzed — this needs deeper investigation into how task/run/create resolves and initializes a Profile when it happens to be the same Profile as the calling/orchestrating Agent (e.g. recursive Profile initialization, working-directory override handling, or credential/session reuse).

Next step: reproduce with a minimal case (Issue Tracker Agent starting a Run with its own Profile ID) and capture the exact error message/stack before proposing a fix.

Filed by the Issue Tracker Agent on user request, for future investigation/fix. No code changes made in this issue.

Start by reading agent-teams/issue-tracker/ and the RFS skill/tasks guidance, then inspect the task/create and task/run/create entry points. Reproduce the self-profile case with a minimal Run and capture the exact error and stack. Done means delegated labels use a short title separate from the goal, and the profile-resolution behavior is understood and addressed.