PR #20437: fix(cli): re-land interrupt_queue drain so finished turns flush stray input

• Repository: NousResearch/hermes-agent
• Author: Tranquil-Flow
• State: open | merged: False
• Link: https://github.com/NousResearch/hermes-agent/pull/20437

Description (problem / solution / changelog)

What does this PR do?

The CLI routes user input typed while _agent_running is True into _interrupt_queue (separate from _pending_input) so the explicit interrupt path can opt to deliver them as a single combined message. That path only drains the queue when busy_input_mode == "interrupt" AND a pending_message was acknowledged.

If the agent's turn finishes naturally (no interrupt fires), any messages typed during the turn stay stuck in _interrupt_queue forever. Subsequent user input can be swallowed behind that stale queue state, making the CLI appear to hang. lunarnexus reports in #20271 that only Ctrl+Z + kill recovers.

The fix restores the post-turn drain that was originally part of #17939. #17666 / #18760 (commit 7696ddc59) explicitly split the drain off as "worth its own review" and never re-landed it; the user-visible regression is that any non-interrupt-mode user typing during a turn is silently dropped.

Implementation extracts the drain to _drain_interrupt_queue_to_pending_input, matching the existing _maybe_continue_goal_after_turn style. process_loop's finally block calls it once per turn after the status-line refresh and before goal continuation (so re-queued user input preempts an auto-continuation prompt). The helper swallows Exception so it can never break the main loop.

Related Issue

Closes #20271

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• cli.py: new _drain_interrupt_queue_to_pending_input helper on HermesCLI; one call from process_loop's finally block.
• tests/cli/test_cli_interrupt_drain_regression.py: 5 cases covering single message, multi-message order preservation, empty-queue noop, falsy-message skip, and exception swallowing.

How to Test

1. pytest tests/cli/test_cli_interrupt_drain_regression.py -v — 5/5 pass.
2. Verify regression: stash the cli.py change and rerun — all 5 fail (helper doesn't exist on main).
3. Wider sweep: pytest tests/cli/ -q — 610/610 pass.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits
• I searched for existing PRs (none — issue-# search and "interrupt_queue" / "pending_input drain" symptom searches return only the parent #17666 / #18760 work)
• My PR contains only changes related to this fix
• I've run pytest tests/cli/ -q and all tests pass
• I've added tests for my changes
• I've tested on my platform: macOS 15 (Darwin 24.6.0), Python 3.14

Documentation & Housekeeping

• I've updated relevant documentation — N/A (internal helper, no public surface)
• I've updated cli-config.yaml.example — N/A
• I've updated CONTRIBUTING.md or AGENTS.md — N/A
• I've considered cross-platform impact — yes, drain is pure-Python queue logic
• I've updated tool descriptions/schemas — N/A

Changed files

• cli.py (modified, +34/-0)
• tests/cli/test_cli_interrupt_drain_regression.py (added, +138/-0)