Decouple Node Execution Plane to Zero-Dependency Go Daemon

[westkevin12]

Description

This PR transitions the Project ORCHID production node execution environment from a Python/Nuitka-based runtime to a hardened, zero-dependency Go/C execution plane, fully resolving the architectural goals in Issue #8.

By replacing the sandboxed Python control plane inside the release-hardened container stage with a dynamically compiled Go binary (orchid-daemon), we have drastically minimized container footprint sizes, cut startup latency to under a second, and fortified the runtime sandbox security using non-privileged user executions.

Closes #8

---

Key Changes

1. Zero-Dependency Go Daemon Execution Core (cmd/orchid-daemon/)

• Ingestion Plane (main.go): Replaced the legacy Python script bootloader with a native Go TCP listener on port 9000 to swallow JSON planning payloads directly.
• Math Parity & Simulation: Re-implemented the parallel STREAM-Triad simulation benchmark in Go, validating 100% bit-for-bit equivalence in cycle counters and event logs matching the legacy Python scheduler.
• Dynamic SIMD Telemetry (matmul_wrapper.go): Deployed a runtime CPUID check linking dynamically to optimized fallback scalar kernels (C-level contiguous loop I-K-J matrices) or native AVX-512 vector assembly.

2. Multi-Stage Container Hardening (Dockerfile)

• Production Stage (release-hardened): Completely removed Python, virtual environments, and compiler tooling. The final image builds on gcr.io/distroless/base-debian12:nonroot and houses only the static orchid-daemon executable.
• Developer Sandbox Stage (developer): Preserved the full Python 3.10 + Astral uv environment for local developers to test raw Python SDK packages (orchid/) or bundle distribution wheel artifacts (make dist).
• Linter Warnings Resolved: Pinned base images to specific tags and wrapped long commands to satisfy strict Hadolint and code quality pipelines.

---

Verification Logs & Proof of Correctness

1. Go Unit Tests (go test -v ./scheduler/...)

=== RUN   TestBankedSchedulerTriad
    scheduler_test.go:129: VERIFY: Mathematical calculations are 100% identical!
    scheduler_test.go:130: Deterministic Serial Cycles: 4925668
    scheduler_test.go:131: Deterministic Parallel Cycles: 1666401
    scheduler_test.go:132: Theoretical Parallel Speedup achieved in Go: 2.956x
--- PASS: TestBankedSchedulerTriad (0.03s)
=== RUN   TestPhysicalNUMAAllocation
--- PASS: TestPhysicalNUMAAllocation (0.00s)
PASS
ok  	ORCHID/scheduler	(cached)

2. Go Native Diagnostics Sweep (./build/orchid-daemon --mode all)

==========================================
Running parallel bank simulation...
workload: A[i] = B[i] + 3 * C[i] | total_elements=16384
latency_cycles: service=100 | compute=1
verification: identical outputs validated | checksum=-1573459

case                                               cycles      speedup        requests/bank
serial_single_memory                              4931584       1.000x              [49152]
parallel_two_memory_role_split                    3293184       1.498x        [32768,16384]
parallel_three_memory_role_split                  1638501       3.010x  [16384,16384,16384]
parallel_two_memory_conflicted_control            4931584       1.000x            [49152,0]

==========================================
Running CPU locality matrix benchmark...
HARDWARE TELEMETRY: AVX-512 not supported. Dispatching to optimized scalar fallback kernel.
VERIFY equal N=512 operations=134217728 cache_flush_bytes=67108864
PAIR 1 order=flat-first flat_sec=0.233877943 locality_sec=0.058633280 speedup=3.989x
...
PAIR 8 order=locality-first flat_sec=0.225897422 locality_sec=0.058633809 speedup=3.853x
FLUSH sink=159383552

3. Hardened Docker Run (docker run --rm orchid-production:latest)

Successfully boots in <0.1s and reproduces timing traces and cycle calculations inside the container environment.