Full hardware UART driver with config-driven per-instance setup.
Non-blocking SendBuffer supports DMA (zero-CPU) and ISR (FIFO-fill)
modes selectable per instance in MCU_UART_cfg.h. Both modes and
blocking SendBufferBlocking invoke a configurable TX-complete
callback (STD_NULL to ignore). All public functions take a u8
instance parameter. Config uses struct-of-arrays pattern for
runtime state, designated-initializer array for per-instance
settings, and value enums in the public header for parity/data
bits/stop bits/async mode. Added hardware_dma to CMake link list.
Mohamed Salem
f50e877ffc