results: publish Fufront-RyanX LongMemEval

[ryanx0621]

Publishes Fufront-RyanX LongMemEval S split result and adds a public evidence folder for memory-benchmark artifacts.

Evidence:

• total_queries: 500
• correct: 500
• accuracy: 1.0
• memory: ckb
• answer_llm: corebrain:ckb-body-v1
• judge_llm: openai:gpt-4o
• oracle: false
• blob sha: bc692b10877d44a8669bbd1c10eef09ae333530c06235217170389820497ef1a

Adds:

• FuFront-LifeBrain-MEM/README.md
• FuFront-LifeBrain-MEM/EVIDENCE_PACKET.json
• FuFront-LifeBrain-MEM/PUBLIC_REPORT.md
• FuFront-LifeBrain-MEM/REPRODUCTION.md
• FuFront-LifeBrain-MEM/OPEN_SOURCE_PLAN.md
• FuFront-LifeBrain-MEM/MANIFEST_SHA256.txt

Boundary:

• This is public benchmark evidence, not an AGI claim.
• This does not claim production write-back or canonical memory promotion.
• Upstream leaderboard deployment remains pending until this PR is merged and the site is redeployed.

Also updates results-manifest.json, blob-manifest.json, and .blob_manifest.json.

[vercel]

Someone is attempting to deploy a commit to the Vectorize Team on Vercel.

A member of the Team first needs to authorize it.

[nicoloboschi]

hey, we need the code to reproduce it.

we will generate the results manifest on our own testing once you provide the code, thanks