fix(update): reinstall editable package with extras

[baketnk]

What does this PR do?

Modifies hermes update to use .[all] to include extras when updating dependencies, to fix cases where extras are added later in development. Currently .[all] is used by install, but not by the update command.

Related Issue

Fixes #1336

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

As above, use .[all] for complete dependency updates. The LLM made a few formatting changes that you can nit at, I glomarize them.

How to Test

Before the discord voice update: install version, then update after voice. voice channel connect fails because of missing deps in venv. (PyNaCl, davey)

Checklist

idk

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: Arch Linux

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

[baketnk]

Reading through the docs again, looks like [all] might not be intended here - might close and think this over.

The problem being, if [all] is pulled by the big dumdum install script, users won't realize there are [extras] for hermes-agent itself. I might close this and rework it to instead print a warning if the voice commands are used without the imports being available.

[teknium1]

Thanks for digging into this. I think the direction is right: using .[all] in hermes update matches how our install scripts and docs already treat Hermes installs, and all does not pull in the RL extra.

A couple changes would make this ready:

1. Please keep installer parity on failure behavior

   • Our install paths currently try -e .[all] first and then fall back to -e . if extras fail.
   • This PR currently makes hermes update fail hard if .[all] fails.
   • Can you switch this to the same try-.[all]-then-fallback-to-. behavior, ideally with a warning that optional features may still be unavailable?

2. Please update both code paths

   • cmd_update() is changed here, but the ZIP fallback update path still reinstalls with -e ..
   • We should keep those paths consistent so Windows / ZIP fallback users get the same dependency behavior.

3. Please adjust the tests to match the intended behavior

   • Right now the new test locks in “extras failure should abort the whole update.”
   • Instead, it should verify we attempt .[all] first, then fall back to . when needed.
   • If practical, add coverage for the ZIP update path too so both update flows stay in sync.

If you want to revise along those lines, I think this becomes mergeable.