fix(power): soft-sleep the GPS receiver during CPU light-sleep

[DatanoiseTV]

GPS only subscribed to notifyDeepSleep, so during light-sleep the
receiver kept drawing ~25 mA on a non-router device — the dominant
drain on devices that spend most of their time in LIGHT_SLEEP.
Subscribe to notifyLightSleep/notifyLightSleepEnd as well, but use
GPS_SOFTSLEEP instead of disable()+enable() since the state machine
survives across the short light-sleep window and full power-cycle
re-acquisition would cost more than it saves.

Split out from #10425 — single-concern PR.

Build verification

pio run -e t-deck-tft succeeds, no new warnings.

Attestations

• I have tested that my proposed changes behave as described — review/static-analysis only, not on-air.
• On-hardware testing requested from community: build-verified t-deck-tft only.

[Copilot]

Pull request overview

This PR reduces GPS power draw during ESP32 CPU light-sleep by subscribing the GPS module to light-sleep enter/exit notifications and transitioning the receiver into a low-power “soft sleep” state across the light-sleep window (instead of fully disabling/enabling the GPS).

Changes:

• Register/unregister GPS observers for notifyLightSleep and notifyLightSleepEnd on ESP32.
• Add ESP32-only GPS callbacks to soft-sleep the receiver on light-sleep entry and restore it on wake.
• Track the pre-light-sleep GPS power state to decide whether to resume GPS_ACTIVE after wake.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/gps/GPS.h	Adds ESP32-only observer members/state for light-sleep enter/exit handling.
src/gps/GPS.cpp	Hooks observers in setup()/destructor and implements prepareLightSleep() / endLightSleep() soft-sleep behavior.

[NomDeTom]

This needs to be reviewed against the main GPS power system. Tying it to CPU sleeping doesn't necessarily make sense.