feat(gallery): add Occamy-1.0 GGUF variants

[localai-org-maint-bot]

Description

Add Occamy-1.0 Q4_K_M and Q8_0 GGUF builds for llama.cpp, with Q8_0 linked as a variant. Both include the F16 vision projector and use the embedded Jinja chat template with an 8K context. Add installation instructions and the publisher's Unicode NFC normalization requirement.

Notes for Reviewers

• Occamy-1.0 is an Apache-2.0 Qwen3.6-35B-A3B derivative for multi-step agent tasks and coding. It appears in the current Hugging Face trending list and is absent from master and existing PRs.
• Use the publisher's GGUF builds. The qwen35moe architecture exists in LocalAI's pinned llama.cpp revision 50631b3d2c569ad8e5c112090cd28570b1268ee0.
• Pin all three unique files to revision e8fe5e28e1b1c1f0cd0a39b85b16b631f17ca14e. Each SHA256 matches both HF LFS metadata and the pinned download's x-linked-etag header.
• Retain the embedded qwen2 tokenizer metadata. Clients must normalize prompt text to NFC; see the linked publisher notes in the docs. The separate experimental MTP head is not configured.
• go test ./core/gallery -count=1 -ginkgo.focus='gallery/index.yaml|gallery variant lint helpers' passes.
• Full go test ./core/gallery -count=1: 392 passed, 3 failed, 9 skipped. All three failures involve forbidden external GitHub/gist fixture downloads. Running those three specs against unchanged master reproduces every failure (request_test.go:22, backends_test.go:208, backends_test.go:213).
• Semantic YAML validation confirms all 1,890 existing entries are unchanged. Both additions have matching model/projector paths, pinned URLs, and hashes. git diff --check passes.
• Model inference was not run locally.

Signed commits

• Yes, I signed my commits. Human DCO sign-off is required; repository policy prohibits AI sign-off.
• Documentation updated (docs/content/) for user-facing changes, or not applicable