docs(seo): improve frontmatter descriptions and add OpenGraph metatags

[itsarbit]

Summary

• Enriched frontmatter description fields across 9 docs pages in docs/main/ so the auto-generated llms.txt (served by Mintlify) contains specific, ArkSim-branded descriptions instead of generic summaries
• Added og:site_name, og:type, og:description, and twitter:card metatags to docs.json for better social sharing and AI discoverability

Why

Mintlify auto-generates llms.txt and llms-full.txt from page frontmatter. Several pages had vague descriptions that didn't mention ArkSim by name, weakening entity clarity for AI answer engines. These changes ensure that when AI models (ChatGPT, Perplexity, Claude, Google AI Overviews) crawl our docs, they get specific, useful descriptions of what each page covers.

Pages updated

Page	Before	After
overview	"Validate, simulate, and harden your agent systems with ArkSim."	Describes ArkSim as an open-source agent testing framework with multi-turn simulation and evaluation
build-scenario	"Define who your simulated user is, what they want, and what they know."	Mentions ArkSim, specifies profile/goal/knowledge
simulate-conversation	"Run scenarios against your agent and collect chats for evaluation."	Mentions ArkSim, clarifies multi-turn conversations
evaluate-conversation	Generic scoring description	Lists actual metrics (helpfulness, coherence, relevance, faithfulness, goal completion)
tool-call-capture	"Capture tool calls from agents that handle tools internally."	Mentions trace receiver and trajectory matching
4 example pages	Generic "walkthrough" phrasing	Specific use case descriptions (compliance metrics, product recommendations, trajectory matching, smart home)

Test plan

• Verify docs.json is valid JSON (pre-commit hook checks this)
• Verify frontmatter YAML is valid in all modified MDX files
• After merge and deploy, check docs.arklex.ai/llms.txt for improved descriptions