Bump to 0.27.1: what the review round found

[Timdebruijn]

Two commits since v0.27.0, both fixes. No new capability, so a patch.

One live bug

g_coredump was read and written across three tasks with no lock. The erase action clears it from
the AsyncTCP task while bridgeInfo() copies its std::string and backtrace from loop(),
rs485Task and AsyncTCP itself — roughly four times a second. That is a use-after-free waiting
for two cores to line up, not a stale read.

Two dormant traps, closed rather than documented away

• The Modbus driver family never took the bus lock, while the dispatcher's comment claimed it
  waited two seconds for one. Harmless while only rs485Task touches a Transport — which is
  exactly why it was worth closing.
• Command outcomes were keyed on a caller-chosen request id alone, so polling one device could
  return another's result. Dispatch was never affected; only the answer an operator reads.

Five documentation corrections

Including one that said enum mode setpoints cannot be expressed, when the profile driver has
implemented them since July.

Verification

1023 native cases, all eleven gates, four board builds. The 1CH build's four warnings are all
inside eModbus, none in src/ — checked again rather than carried over from the last note.

[Copilot]

Pull request overview

Updates the firmware patch version to 0.27.1 and records the release-notes narrative in src/main.cpp alongside prior version notes, aligning the in-source version constants with the stated patch release.

Changes:

• Bump HELIOGRAPH_VERSION_PATCH from 0 to 1 (0.27.1).
• Add a detailed 0.27.1 release-notes comment describing the fixes found during review.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.