Bump mlx to version 2.21.2

[LeonNissen]

Bumps mlx to version 2.21.2

♻️ Current situation & Problem

The MLX library is currently 1.18.1, and can be bumped to 2.x to support newer models.

⚙️ Release Notes

Updates the MLX library to 2.21.2 and migrates code for breaking changes.

📝 Code of Conduct & Contributing Guidelines

By submitting creating this pull request, you agree to follow our Code of Conduct and Contributing Guidelines:

• I agree to follow the Code of Conduct and Contributing Guidelines.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 85 lines in your changes missing coverage. Please review.

Project coverage is 38.34%. Comparing base (26b1e07) to head (7a1e836).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...urces/SpeziLLMLocal/LLMLocalSession+Generate.swift	0.00%	82 Missing ⚠️
Sources/SpeziLLMLocal/LLMLocalSession+Setup.swift	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #94      +/-   ##
==========================================
- Coverage   38.57%   38.34%   -0.23%     
==========================================
  Files          64       64              
  Lines        2331     2345      +14     
==========================================
  Hits          899      899              
- Misses       1432     1446      +14     

Files with missing lines	Coverage Δ
Sources/SpeziLLMLocal/LLMLocalSchema.swift	0.00% <ø> (ø)
Sources/SpeziLLMLocal/LLMLocalSession.swift	0.00% <ø> (ø)
Sources/SpeziLLMLocal/LLMLocalSession+Setup.swift	0.00% <0.00%> (ø)
...urces/SpeziLLMLocal/LLMLocalSession+Generate.swift	0.00% <0.00%> (ø)

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 26b1e07...7a1e836. Read the comment docs.

[PSchmiedmayer]

Very cool; thank you for bumping this so quickly.

Happy to see it merged once all CI elements are passing and the existing warnings are addressed 🚀

[vishnuravi]

Sorry @LeonNissen I was playing around with this branch and pushed a commit by mistake. I reverted it. Didn't mean to step on your toes.

[LeonNissen]

Sorry @LeonNissen I was playing around with this branch and pushed a commit by mistake. I reverted it. Didn't mean to step on your toes.

@vishnuravi No worries, feel free to play around with it, I am currently working on a different branch anyways 😊

[vishnuravi]

Sorry @LeonNissen I was playing around with this branch and pushed a commit by mistake. I reverted it. Didn't mean to step on your toes.

@vishnuravi No worries, feel free to play around with it, I am currently working on a different branch anyways 😊

Oh ok, cool, I will continue to work on this branch then.

[vishnuravi]

The Linkspector check is failing with a 403 when attempting to verify links to the OpenAI documentation that are valid and reachable from a regular web browser. This suggests that openAI is likely blocking GitHub IPs. I would make this check optional until these false positives are solved.

[PSchmiedmayer]

That's unfortunate.
@vishnuravi, what about adding a config file in the repo and adding the openai web page to the ignore pattern or just exclude the very specific link in the ignore pattern?

[vishnuravi]

The documentation for Keychain Access Groups entitlement and Increase Memory Limit entitlement appear to be gone from the Apple website. Is this still relevant in the README? @philippzagar @LeonNissen

[vishnuravi]

@PSchmiedmayer addressed all your comments with a re-factor. Link check is also now passing. @LeonNissen and I tested on devices. Please take a look and we can merge if you have no other feedback. Thanks.

[philippzagar]

The documentation for Keychain Access Groups entitlement and Increase Memory Limit entitlement appear to be gone from the Apple website. Is this still relevant in the README? @philippzagar @LeonNissen

Interesting that Apple just removed these entitlements.. @LeonNissen As you added these docs (I think), are they still beneficial in local SpeziLLM? If yes, is there an alternative?

[philippzagar]

Thank you for upgrading to the next major MLX version @LeonNissen and dealing with all the breaking changes! 🚀
Just had minor follow up questions.

[vishnuravi]

Thank you for upgrading to the next major MLX version @LeonNissen and dealing with all the breaking changes! 🚀 Just had minor follow up questions.

Thank you for the review @philippzagar! Look forward to merging this and then getting the new DeepSeek models included via #97!

[vishnuravi]

Rebased and fixed unsigned commits.