docs: bring the README up to date with the deployed node

[bazauto]

The README still described the repo as it was before the node split, so several parts of it were wrong. This brings it into line with what is actually in the repo and on the bench. It only changes docs.

What was wrong, and what changed:

• Status still listed Restructure into two nodes with a shared library #1 to End-to-end bring-up: one current-sensing and one IR sensor #3 as in progress, but all three are closed and the node is deployed end to end. Status now lists what is still open: A broken sensor wire reads as clear track, not occupied (CS fix under test; IR open) #9, IR beam at Goods Shed oscillates with a stationary loco over it #10, Enable TLS on the broker, with server-certificate verification on every client #25 and RFID node: restructure only, pending hardware and orchestration#39 #4, plus point feedback. Point feedback is built (Point position feedback node on board 3 (0x22) #15) but publishes nothing until a feedback source is chosen.
• The layout section said src/apps/io-node/ did not exist and that a root main.py was still running. Both were false. The section now shows src/apps/io-node/ and scripts/deploy.sh.
• The Cobalt row said point position comes from the S2 changeover. That's wrong since docs: the Cobalt has no free contact for point feedback #22: S2 powers the frog, and the feedback source is undecided.
• Hardware now lists three MCP23017s, with board 3 at 0x22, and says sensor polarity is set per sensor.
• Deploying now shows the command itself, says the node needs its broker credentials (Broker accepts loopback connections only, so no node can publish #7, docs/broker-auth.md), and points to the LED codes.
• The repo list now links block-detection, the board that fixes A broken sensor wire reads as clear track, not occupied (CS fix under test; IR open) #9.

The licence section is unchanged.

🤖 Generated with Claude Code