feat: improve thought segments syntax extraction

[giladgd]

Description of change

• feat: improve thought segment syntax extraction
• feat: expose the model's architecture directly on the model instance
• feat: expose download speed and ETA for a model downloader
• feat: export gguf file quant names
• fix: adapt to breaking llama.cpp changes
• fix(LlamaContextSequence): make .dispose() return a promise
• fix: Qwen chat wrapper auto thought segment opening
• fix: optimize checkpoints with auto opening thought segments
• fix: map GgmlType configs directly from native code
• fix: simulator model dispose while in use race conditions
• fix: safer cleanup flows
• fix: bugs
• refactor: Vulkan device memory readings
• docs: add type links

Fixes #634
Fixes #637 (native object disposal race conditions)
Fixes #629 (native object disposal race conditions)
Fixes #623

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)

[JoshuaWise]

Okay! This PR is perfect. You probably got a bunch of notifications in your email from previous comments I made which I've deleted because I realized they were wrong. I didn't realize additionalRenderParameters.enable_thinking was replaced by reasoning in this PR, so my testing was off. Ignore all my previous comments. This PR perfectly addresses #634 and I'm stupid. Thank you for the fix!

[ido-pluto]

LGTM

[github-actions]

🎉 This PR is included in version 3.20.0 🎉

The release is available on:

• v3.20.0
• npm package (@latest dist-tag)
• GitHub release
• v3.20.0

Your semantic-release bot 📦🚀