Use SPI.transfer(wbuf,rbuf,count) also for read

[jayzakk]

While doing performance tests, I noticed that the special SPI.transfer(writebuf,readbuf,count) is only called for write, but not for read. As this massively speeds up data throughput on supported MCUs, please accept this PR.

Here: Optimized performance on LGT8F-MCUs (Nano/Atmega328p replacement), up to 14.1Mbit/s receive and 10.7Mbit/s send. https://github.com/dbuezas/lgt8fx/wiki/Libraries-support

[CLAassistant]

All committers have signed the CLA.