docs: align roadmap and repository story with current SPARK

[adrianwedd]

Summary

• replace the generic robotics/RL roadmap with the issue-linked trajectory of the persistent robot SPARK actually is now
• replace the stale 831-line PiCar-X Hacking README with a concise current architecture overview
• render the homepage roadmap from the same current priorities instead of the old generic robotics-future list
• add editorial regression tests so README/roadmap cannot silently drift back to the pre-resident architecture

Roadmap emphasis

Now: companion/admin split, explainable action traces, pseudo-agency cleanup, GPIO lease completion, quiet/audio semantics, perception/retention boundaries, wake-memory/IO pressure.

Next: persistent spatial memory, lightweight fossil/sim replay, predictive health, teach-with-SPARK and physical play.

Later: docking/energy awareness, long-horizon spatial memory, bounded self-maintenance, more local cognition, portable persistent state.

Explicitly de-prioritises generic RL dream buffers, fleet policy sharing, multi-car demos and central map/log platforms without a concrete one-robot need.

Additional audit finding

The sweep found remaining stale canonical/public copy outside this PR: CLAUDE.md still documents the deleted claude-voice-bridge / three-backend voice-loop architecture, and lower homepage Brain/FAQ copy still contains some pre-M5 framing. Filed as #254 with exact acceptance criteria rather than broadening this PR into a risky rewrite of the operational constitution.

Verification

• tests/test_site_story.py now pins the repository README, canonical roadmap and rendered homepage roadmap alongside the Tell the current SPARK story #253 public-story invariants.
• No production/backend code changed.

[cloudflare-workers-and-pages]

Deploying spark with    Cloudflare Pages

Latest commit:	b9876eb
Status:	✅  Deploy successful!
Preview URL:	https://47a324d7.spark-e11.pages.dev
Branch Preview URL:	https://docs-current-roadmap-and-rea.spark-e11.pages.dev

View logs