Fix pinMode pulls not driven on ESP-IDF v6 (SOC_GPIO_SUPPORT_RTC_INDEPENDENT removed)

[ainyan03]

ESP-IDF v6 removed the SOC_GPIO_SUPPORT_RTC_INDEPENDENT soc capability macro. pinMode() treated the missing macro as 0, which routes RTC-capable pins through the RTC-domain pull path that is only appropriate for the plain ESP32. On ESP32-S2/S3 (and other chips whose digital pads are independent of the RTC domain) that path does not drive the pad pulls at all: input_pullup left the IO_MUX FUN_PU bit clear and the line never rose. Anything relying on pinMode pulls without an external resistor — e.g. the software I2C probing used by board autodetection — silently loses its pull-ups on ESP-IDF v6.

Fix

Map the capability to an internal LGFX_GPIO_RTC_INDEPENDENT macro: use the SoC-provided value when available (ESP-IDF v5 and earlier), otherwise derive it from the build target — only the plain ESP32 needs the RTC path. This mirrors the internal GPIO_RTCIO_ARE_INDEPENDENT logic ESP-IDF v6 itself uses, and avoids re-defining a macro that ESP-IDF has officially retired.

Verification

Measured on M5StopWatch (ESP32-S3) with a floating RTC-capable pin (GPIO4), applying input_pullup / input_pulldown / input and reading back the input level and the IO_MUX pull bits:

	ESP-IDF v5.5.4	v6.0.1 before	v6.0.1 after
input_pullup	high, FUN_PU=1	low, FUN_PU=0	high, FUN_PU=1

• No regression on v5.5.4 (macro still provided by the SoC caps → same value as before)
• No behavior change for the plain ESP32 (derivation yields 0, matching the previous state)
• Builds: Arduino-ESP32 2.x / 3.x (esp32, esp32s3), ESP-IDF v6.0.1 (esp32, esp32s2, esp32s3), v5.5.4 (esp32s3)
• CI on the fork: build workflows green