homepage: capability equivalence on the AI cards

Every fleet model card (and the #models docs section) now states, in
plain family-facing terms, what the model is equivalent to:
frank-large ~ Claude Sonnet 5 / OpenAI-Luna, frank-primary ~ Claude
Opus 4.8 / OpenAI-Terra (both Summer 2026 frontier), frank-secondary ~
last year (2025) frontier. Rendered as a new details.equiv line
(desc-equiv, styled in the head color, bare-mode clean); pinned in
homepage_config.bats.

shared/homepage/services.json

@@ -269,6 +269,7 @@
     "section": "AI",
     "details": {
       "role": "Deep reasoning model",
+      "equiv": "Equivalent to Claude Sonnet 5 / OpenAI-Luna (Summer 2026 frontier intelligence).",
       "spec": "Qwen3.8-Flash-Next v2 Swift1.5 abliterated (halogen flash-server, ~62 GiB) — 200k-token context, two conversations at a time; sees images. Speculative decoding always on.",
       "inputs": ["text", "image"],
       "use": "Cybersecurity and debugging. Requires Frank's permission."
@@ -286,6 +287,7 @@
     "section": "AI",
     "details": {
       "role": "Primary model",
+      "equiv": "Equivalent to Claude Opus 4.8 / OpenAI-Terra (Summer 2026 frontier intelligence).",
       "spec": "Qwen3.8-Flash-Next at native FP8 precision (radiance, 173 GiB) — 200k-token context, two conversations; sees images. Runs on both R9700s with tiered expert memory.",
       "inputs": ["text", "image"],
       "use": "Software engineering, system administration, and instruction tasks."
@@ -303,6 +305,7 @@
     "section": "AI",
     "details": {
       "role": "Fast utility model",
+      "equiv": "Equivalent to last year's frontier level (2025).",
       "spec": "Qwen3.5-9B-Uncensored (Q6_K, 6.9 GiB) on llama.cpp CUDA — two 200k-token conversations at a time; sees images; ~95 tok/s decode on the RTX 3090.",
       "inputs": ["text", "image"],
       "use": "Agentic search, RAG, research subagents, translation, roleplay, and summarization — any small task that doesn't need much intelligence."