fix(cli): enable editing in custom model prompt

[helix4u]

What does this PR do?

Adds normal line editing to the custom model-name prompt. On an interactive terminal, the prompt now uses Hermes' existing prompt_toolkit dependency, so Left/Right and Ctrl+A/Ctrl+E edit the entered model name instead of printing raw escape bytes.

The helper is intentionally narrow: redirected input/output keeps the existing built-in input() path, imports stay lazy, and password or masked-secret prompts are unchanged.

The root cause was the transition from the curses model picker to a bare input("Enter model name: ") call. A scripted Python process does not initialize a readline-style editor for that call, so affected terminals pass cursor keys through as visible VT sequences.

Related Issue

Reported in the Discord support thread arrow keys are not correctly processed by CLI wizard.

Related open PRs #83045 and #40458 affect the curses picker itself, but neither changes the subsequent custom model-name prompt covered here.

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

• hermes_cli/cli_output.py: add a TTY-aware non-secret line-input helper backed by prompt_toolkit, with the existing built-in input() behavior retained for redirected streams or an unavailable dependency.
• hermes_cli/auth.py: use that helper in both custom model-name branches, including the numbered fallback.
• tests/hermes_cli/test_cli_output.py: cover Left/Right and Ctrl+A/Ctrl+E editing through a real prompt-toolkit pipe, plus the redirected-input fallback.
• tests/hermes_cli/test_terminal_menu_fallbacks.py: verify both model-selector paths use the line editor.

How to Test

1. Run hermes model and select a provider with a model picker.
2. Choose Enter custom model name.
3. Type a model name, move with Left/Right, and use Ctrl+A/Ctrl+E. The cursor should move and edits should be inserted at its position without visible escape bytes.
4. Run tests/hermes_cli/test_cli_output.py and tests/hermes_cli/test_terminal_menu_fallbacks.py.

Local validation:

• ruff check on all four changed files: passed
• ruff format --check on the newly formatted helper and helper-test files: passed
• git diff --check: passed
• Focused Python tests were added but not run locally; GitHub CI is expected to run them.

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

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) - N/A; no documented workflow changes
• I've updated cli-config.yaml.example if I added/changed config keys - N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows - N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide
• I've updated tool descriptions/schemas if I changed tool behavior - N/A

Screenshots / Logs

Not applicable. The regression is exercised with prompt-toolkit pipe input so the terminal key sequences remain deterministic in CI.

[teknium1]

Merged via PR #90345 with your commit cherry-picked onto current main, authorship preserved in git log. We widened your line_input() fix to all 46 bare input() free-text prompt sites across the setup wizards plus the shared prompt() helpers. Thanks!