fix: Add graceful embedding fallback to store_reasoning_memory

[kevinkre]

• Create zero-vector fallback when embeddings fail
• Continue to text-only storage instead of failing
• Add usingEmbedding and storageMode indicators to response
• Matches graceful fallback behavior of cipher_memory_search
• Enables memory storage even when embedding providers fail

[RyanNg1403]

@codex review this PR

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Breezy!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[RyanNg1403]

@kevinkre thanks for your PR, can you double-check to make sure all tests are passed before I merge this PR?

[PROJECT-OCULUS]

Test Confirmation ✅

I've run the full test suite on the latest code including this PR's changes:

Test Files  104 passed (104)
Tests       2057 passed (2057)
Start at    17:50:26
Duration    53.73s

All 2057 tests pass. The fallback behavior for embeddings is working correctly and maintains backward compatibility.

Ready to merge!

[RyanNg1403]

Hey @kevinkre, closing this PR due to inactivity.

We asked you to verify that all tests pass back in October 2025, but we never heard back. It's been about 5 months without a response.

The fix itself (graceful embedding fallback to zero-vector text-only storage) is a good idea — if you'd like to revisit it, feel free to open a fresh PR rebased from current main. Thanks for the contribution.