Power: output setters return bool, add getPowerOutputCaps()

[ainyan03]

Summary

Power_Class::setExtOutput, setUsbOutput, setExtPortBusConfig and setVibration returned void and discarded the result of the I2C / IO expander / PMIC operations behind them, so a caller could not tell whether the external 5 V, the USB output or the vibration motor were actually switched. This PR makes the four setters report whether the requested state was reached and adds getPowerOutputCaps() to tell "this model has no such path" apart from "the request failed", in the same shape as setChargeCurrent / getChargeControlCaps().

API

API	Before	After
setExtOutput(enable, port_mask)	void	bool – true when every selected port reached the requested state
setUsbOutput(enable)	void	bool
setExtPortBusConfig(config)	void	bool
setVibration(level)	void	bool
getPowerOutputCaps()	–	uint8_t bitmask of power_output_capability_t (cap_set_ext_output / cap_set_usb_output / cap_set_ext_port_bus / cap_set_vibration); 0 before M5.begin() completes and on the PC build
AXP192_Class::setEXTEN / setGPIO* / setLDO*, AXP2101_Class::set*LDO*	void	bool
AXP192_Class::getBatteryLevel(int8_t*), INA3221_Class::getShuntMilliVoltage(ch, int32_t*), M5PM1_Class::getGPIOOutputLatch(pin, bool*)	–	checked variants (I2C result separated from the value)

Behaviour

• false covers: a model without the path (see getPowerOutputCaps()), an I2C failure, a protection check that cancelled the request (Core2 / Tough on a low battery, the CoreS3 family without a battery), the AW9523 mutex not being available or the request being superseded on CoreS3, an unconfirmed PMIC identity (same rule as the charge setters), and a port_mask that selects none of the model's ports (ext_none on every model).
• Where port_mask selects several ports (M5Station, PowerHub, Tab5 family, CoreP4X, ChainCaptain) every selected port is still attempted; false means at least one did not take effect. Models with a single output ignore the mask.
• Core2 / Tough: the readings used by the protection check are now checked reads. A reading that cannot be obtained refuses the enable request instead of passing the check with a folded-to-zero value.
• Disable requests stay best effort: the AXP192 / AXP2101 rail setters and the Core2 / Tough EXTEN write attempt every step after an earlier failure, so a failed voltage write no longer skips the enable bit.
• setVibration: a non-zero level is clamped up to the lowest voltage the rail can produce (AXP192 LDO3 1800 mV, AXP2101 DLDO1 500 mV). Levels below it used to switch the rail off while the call looked successful.
• Core2 v1.1 / Tough with AXP2101: when the INA3221 did not initialise, the current-direction check is omitted as before; a detected INA3221 that fails to read refuses the enable request (logged).
• ChainCaptain: the boost output follows the port latches only when a port was selected; when the latch cannot be read the boost is left untouched and the call returns false (it used to switch the boost off).
• AXP192 / AXP2101 rail setters: a disable request reports only whether the enable bit was cleared; enable reports both the voltage and the enable write.
• CoreS3 family: the AW9523 output helper reports each write; a request that could not take the mutex or was superseded by a later request returns false (it did not take effect).

Breaking changes

• Return types changed (void → bool). Calls that ignore the result still compile. A return type is not part of the mangled name: rebuild everything that links against the library.
• setExtOutput(enable, ext_none) returns false and does nothing on every model (it used to switch the single output on models that ignore the mask).
• setVibration(1..109) on a Core2 with AXP192 (and setVibration(1) with AXP2101) now drives the motor at the rail minimum instead of switching it off.
• The deprecated setExtPower wrapper stays void.

Not changed (separate issue)

• CoreS3 family: setExtOutput(false) clears BOOST_EN without looking at USB_EN, which stops an enabled USB output. This condition has been in place since the first CoreS3 implementation and is out of scope here.

Verification

Build: ESP32 / ESP32-S3 (×2) / C6 / C5 / P4 Arduino, ESP32 ESP-IDF 6.0, native (PC build).

Board	Path	Checked
Core BASIC	no PMIC	caps = 0; every setter false
Tough	AXP192	caps = ext; on/off with read-back, ext_none false, usb / vib / bus false
ToughC5	M5PM1	caps = ext; on/off with read-back, ext_none false
PowerHub	STM32 front-end	caps = ext + bus; on/off per port (USB / PA) with read-back, non-existent port false, bus config accepted
Core2 v1.3	AXP2101 + INA3221	caps = ext + vib; ext on/off with read-back, vib 1 / 0 true
Core2 v1.1	AXP2101 + INA3221	caps = ext + vib; ext on with read-back (ext off drops the USB console on this unit, same with develop)
CoreS3 (DIN base)	AXP2101 + AW9523	caps = ext + usb; ext on/off with read-back (twice), usb off, ext_none false; the no-battery protection cancel reports false