[iBobbyTS]

Problem

FanControl.resetAuto() discarded every writeKey result and logged "Reset to Apple defaults" unconditionally. When a write is rejected — the common case being a non-root caller, where AppleSMC allows reads but the firmware rejects fan-mode writes — the fans stay latched in manual mode while every caller believes the reset succeeded. Two callers are directly affected today: the SIGTERM shutdown path logs "released fans to auto" from this call's result, and the watchdog / .auto paths rely on the throw to tell the client the truth.

This is the same diagnosis as #30 — that PR is the right direction and this builds on it.

Fix

• Every fan is attempted before throwing, so one uncooperative key cannot strand the remaining fans in manual (mode/Ftst part of core: resetAuto no longer reports success when its SMC writes were rejected #30).
• Beyond core: resetAuto no longer reports success when its SMC writes were rejected #30: fans whose AUTO-mode write failed stay in manual mode, and writing their target to 0 anyway would command a manual 0 RPM — a stopped, unsupervised fan. They keep their last (spinning) target instead, and the throw makes the caller retry the whole reset.
• Target-write and Ftst failures are also reported (first failing key wins).

Testing

swift build + full swift test pass (152 tests). The change is in the SMC write path itself, which (like #30) has no hardware-free seam to unit-test; behavior is covered by review of the write sequence.