[Request] let startWrite() & writeAckPayload() return a bool

[2bndy5]

This is a follow up to #678 from tracking the return values for write_register() & write_payload().
Most of the function calls were ignoring the returned status byte from write_register() & write_payload(). Since I proposed that all SPI transacting functions save the status byte to the private member status, the write_payload() function should also save the status byte to the private member status.

Furthermore, I propose that startWrite() & writeAckPayload() instead return a boolean describing if the payload was able to write to the TX FIFO. The status byte fetched over MISO during the SPI transaction can tell us if the nRF24L01 has ignored the new payload due to no space in the TX FIFO. Thus true means payload was written, and false means the payload was ignored.

Currently startWrite() & writeAckPayload() both return void, but they could be more helpful if they returned a boolean about the success of writing the payload to the TX FIFO.

New definition for startWrite():

bool RF24::startWrite(const void* buf, uint8_t len, const bool multicast)
{
    // Send the payload
    write_payload(buf, len, multicast ? W_TX_PAYLOAD_NO_ACK : W_TX_PAYLOAD);
    ce(HIGH);
    #if !defined(F_CPU) || F_CPU > 20000000
    delayMicroseconds(10);
    #endif
    ce(LOW);
    return !(status & _BV(TX_FULL)); // was the payload ignored or did it get clocked properly to the TX FIFO?
}

Notice the function still performs as normally, but also returns information about the TX FIFO for better context to the MCU. writeAckPayload() will need some additional tweaking as it seems to access the SPI bus directly (though it wouldn't be difficult).

Additional Context

write() could use the status byte to flush the TX FIFO if it is full since write() is intended to only use the top level of the TX FIFO per call, but this seems like an added limitation (similar to the detriment that txStandBy() flushes all payloads from the FIFO on an individual failed transmission).