[BOUNTY #16257] Test corpus + fuzz harness for node x86/device arch classification

[CHY9213]

Summary

This PR adds a comprehensive test corpus and fuzz harness for the node's x86/device architecture classification path, as specified in bounty #16257.

Deliverables

Real-world corpus tests (47 samples total)

Corpus File	Samples	Coverage
vintage_x86_samples.json	9	i486, Pentium, AMD K6, PII/PIII, Pentium 4, Athlon XP/64, Phenom II
modern_x86_samples.json	10	Sandy Bridge through Raptor Lake, Xeon servers, Ryzen/EPYC, FX
arm_sbc_samples.json	4	RPi4/5, Orange Pi 5, Jetson Nano
powerpc_samples.json	4	G3/G4/G5, AmigaOne
vm_qemu_samples.json	6	QEMU/KVM, VMware, VirtualBox, WSL2, Docker
adversarial_samples.json	14	Spoof payloads (see below)

Adversarial cases (14 samples)

• Empty CPU brand + no machine field
• ARM platform.machine() with x86 brand claim
• cpu_family 6 + Pentium III injected brand string
• Legacy key-name trap (entropy profile hash reads legacy key names -> 7 of 9 fields are dead defaults on the shipping payload (and the naive fix has a downside) #7991): cpu_brand empty but brand_string/model_name populated
• Unicode, oversized, null-byte fields
• Cross-arch spoofs (PPC brand + x86 features, Apple M1 + SSE)
• Malformed payloads (list features, float family, empty)

Fuzzing (hypothesis property-based)

• Random string inputs never crash detect_cpu_architecture()
• Random string + loyalty values never crash calculate_antiquity_multiplier()
• No multiplier > 1.0 without vintage evidence (spoof guard)
• Dict-shaped inputs to arch_cross_validation.validate_arch_consistency() are safe
• Short/truncated ASCII strings handled
• Seeded fuzz from corpus brands

CI compatibility

• Added hypothesis>=6.0.0 to requirements.txt
• Tests run under pytest tests/ via existing CI
• Fuzz uses bounded example budget (max_examples=200)

Acceptance criteria

• Corpus + adversarial tests pass (or document failures as xfail)
• Fuzz harness runs in CI with bounded time
• entropy profile hash reads legacy key names -> 7 of 9 fields are dead defaults on the shipping payload (and the naive fix has a downside) #7991 legacy key-name case explicitly covered
• No behavioral change to classification logic

Closes Scottcjn/rustchain-bounties#16257

[Scottcjn]

Thank you for this — it's genuine work and I want to be straight with you about what happened. Bounty #16257 is decided by the first mergeable PR rule. Yours was submitted first, but it's currently in a CONFLICTING state (merge conflict with main), so it wasn't mergeable when the bounty was settled — and guoqiangliu's #8027 was clean and merged. So #16257 went to them.

That's not a knock on your code. If you rebase this onto current main and it's clean, ping me — I keep an eye out for solid contributors, and there's a steady stream of open bounties where being first-and-clean wins. Genuinely hope you stick around. — Sophia, Elyan Labs