0.2.26

Fixes

• M5Stack Basic and other boards with the ESP32 internal DAC speaker: every tone ended with a short drop to the DAC zero level followed by a jump back to the bias, heard as crackling. The DMA is now fed at the hold level while waiting for new data. A new sound requested during that hold starts about one DMA ring later. (#371, #391)
• CoreS3 / CoreS3 SE / Core2 v1.1: on units whose AXP2101 EFUSE leaves the TS pin as a charger input, the battery never charged (the unconnected TS reads as a cold battery). TS is now set as an external input. (#387, #388)
• NanoC6 / NanoH2: M5.Led never lit; the RGB LED supply switch (NanoC6 GPIO19, NanoH2 GPIO10) was not turned on. It is switched on at the first LED access and off by release(). (#389)
• Capsule: GPIO13 / 15 were left as outputs driving High after board detection (the I2C STOP on the Grove port), so input-only drivers such as RMT RX could not see the signal. The pins are restored after the sequence. (#390, #393)
• CoreS3 / CoreS3 SE / Core2 v1.1: the AXP2101 charge voltage register is not always at its datasheet default at power-on (a unit read a reserved value). begin() now sets 4.2 V explicitly. (#387, #392)
• CoreS3 / CoreS3 SE: setExtOutput(true) precharges the 5 V bus with short pulses before enabling it, so that the inrush into the empty bus does not power the board off on a weak USB supply with no battery. (#381)
• ToughC5: setChargeCurrent() selected the high charge step as 830 mA; the hardware step is 650 mA, so requests of 650 to 829 mA chose 180 mA instead. (#395)