#6690: UX: clearer /queue and /subagents output

Small UX-only change for chat commands: - /queue (no args) now shows a multi-line status card including followup queue depth and whether session settings override defaults. - /subagents list output is split into two lines per run (status/label then timing + ids) and uses uppercase status for scanability. Testing: pnpm vitest run src/process/command-queue.test.ts src/auto-reply/reply/subagents-utils.test.ts <!-- greptile_comment --> <h2>Greptile Overview</h2> <h3>Greptile Summary</h3> This PR tweaks the UX for two chat commands: - `/subagents list` output is reformatted into a header plus two lines per run (status/label, then timing + IDs) and uppercases status for scanability. - `/queue` status output (no args) becomes a small multi-line “status card” that includes followup queue depth and whether session-level queue settings override defaults. Changes are localized to the auto-reply command handling layer (`src/auto-reply/reply/*`) and do not alter queue/subagent mechanics themselves, only how the current state is rendered in replies. <h3>Confidence Score: 3/5</h3> - Generally safe to merge, but `/queue` status may show misleading values in common cases. - The PR is mostly presentation-only, but the new `/queue` status card introduces two behavioral accuracy risks: `followup_depth` appears keyed off `sessionId` rather than the string key used by followup queues in other paths, and the new `overrides` detector can miss explicit zero values. If these assumptions are wrong, users will see incorrect status output. - src/auto-reply/reply/directive-handling.queue-validation.ts <!-- greptile_other_comments_section --> _{(3/5) Reply to the agent's comments like "Can you suggest a fix for this @greptileai?" or ask follow-up questions!} <!-- /greptile_comment -->