What does this PR do?

This PR reduces visible “waterfall” redraw during classic CLI terminal resize/redraw recovery.

Before this change, _replay_output_history() replayed recent output history one rendered line at a time. During terminal resize recovery, that meant every history line was sent through prompt_toolkit separately, causing the previous conversation output to visibly reappear line-by-line. This looked like the conversation was being streamed from the beginning again, and it also forced prompt_toolkit to restore/redraw the prompt chrome repeatedly.

This change keeps the existing output-history data model and recording semantics unchanged, but changes the replay transport: replay lines are rendered first, joined into a single ANSI payload, and printed once. That preserves the existing resize/redraw recovery behavior while avoiding the repeated per-line print/redraw loop.

This is intentionally scoped to the most disruptive UX symptom: output-history replay waterfall during classic CLI resize/redraw recovery. It does not attempt to solve full width-aware semantic reflow, ghost-line cleanup, live assistant streaming smoothness, toolbar flicker, or Ink TUI (--tui) resize behavior.

Related Issue

Addresses the output-history rapid replay symptom discussed in:

• fix(cli): skip output-history replay on terminal resize #21972 — describes _OUTPUT_HISTORY being replayed after resize, making recent output look like it is “re-streaming from the beginning”.
• Terminal resize causes status bar duplication and blank line flooding on macOS #19280 — includes follow-up context that _replay_output_history() rapidly reprints recent output lines after SIGWINCH.

Related umbrella tracking issue:

• feat: smooth terminal reflow on window resize (parity with opencode/Claude Code) #24164 — broader smooth terminal reflow/parity tracking issue. This PR only handles one replay-transport symptom and should not close the umbrella issue.

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• cli.py

  • Batch _replay_output_history() output into one joined ANSI payload instead of printing each replay line separately.
  • Preserve the existing output-history entries and callable-entry rendering behavior.
  • Keep replay output disabled while replaying, as before, so replayed content is not re-recorded into history.

• tests/cli/test_cprint_bg_thread.py

  • Update the replay test expectation from multiple per-line prints to one batched print.
  • Add coverage for batching rendered output-history lines into a single prompt_toolkit print call.

How to Test

1. Run the targeted regression tests:

   uv run pytest tests/cli/test_cprint_bg_thread.py tests/cli/test_resume_display.py -q -o 'addopts='

   Expected result from local verification:

   49 passed in 1.90s
   

2. Manually reproduce the original UX issue in the classic CLI:

   uv run hermes

   Then generate enough output to fill several lines, and resize the terminal window wider/narrower or trigger a redraw/recovery path.

3. Verify that recent history is restored as a single redraw rather than visibly replaying line-by-line like a waterfall.

4. Confirm that normal live assistant streaming is unchanged. This PR only changes output-history replay during resize/redraw recovery; it does not batch normal model streaming output.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS 15.7.3

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

Targeted tests run locally:

uv run pytest tests/cli/test_cprint_bg_thread.py tests/cli/test_resume_display.py -q -o 'addopts='
.................................................                        [100%]
49 passed in 1.90s

Local manual verification:

• In classic CLI mode, terminal resize after long output restores recent output in one redraw instead of line-by-line waterfall replay.
• Normal live assistant response streaming remains separate and unchanged.