feat: automatic checkpoints for models that need it

[giladgd]

Description of change

• feat: automatic checkpoints for models that need it (such as Qwen 3.5 due to its hybrid architecture)
• feat(QwenChatWrapper): Qwen 3.5 support
• feat(inspect gpu command): detect and report missing prebuilt binary modules and custom npm registry
• feat: initial disk cache dir option for future optimizations (disabled for now)
• fix: Qwen 3.5 memory estimation
• fix: grammar use with HarmonyChatWrapper
• fix: add mistral think segment detection
• fix: compress excessively long segments from the current response on context shift instead of throwing an error
• fix: default thinking budget to 75% of the context size to prevent low-quality responses
• fix: bugs
• test: Qwen 3.5 checkpoint use
• chore: update modules

Fixes #571
Fixes #575
Fixes #577

Pull-Request Checklist

• Code is up-to-date with the master branch
• npm run format to apply eslint formatting
• npm run test passes with this change
• This pull request links relevant issues as Fixes #0000
• There are new or updated unit tests validating the change
• Documentation has been updated to reflect this change
• The new commits and pull request title follow conventions explained in pull request guidelines (PRs that do not follow this convention will not be merged)

[ido-pluto]

LGTM

[github-actions]

🎉 This PR is included in version 3.18.0 🎉

The release is available on:

• v3.18.0
• npm package (@latest dist-tag)
• GitHub release
• v3.18.0

Your semantic-release bot 📦🚀