T4 - Clear out TX FIFO in RequestFrom and endTransmission

This appears to resolve the hang in the BNO055 test cases.

Basically I detect if the FIFO queue is not empty when you start these two functions and clear it out if it does.  Also NEED to then clear out the FEF status bit after you clear it.

You might want to see if there is a better way.  But @mjs513 said it also resolved his issue with MPL3115

T4 Wire1 has alternate pins - setSCL setSDA implemented

T4 has alternate wire pins on the SDCard pins, which may come in handy in some cases.

So converted the pin initialization stuff to be close to what I did in hardwareSerial, and created a pin definition structure with pin numbers, and then mapped those pins to underlying IMXRT data...

Implemented setSCL and setSDA.

Tested with  modified version of Scanner.ino, and found prop shield devices on Wire, Wire1, Wire2 and then Wire1 on alternate pins...

Added APDS-9960 lux sensor at 0x39

Used in the sparkfun apds-9960 lux rgb proximity sensor boards and clones.

fixes #12

F_BUS 128MHz

Make TwoWire constructor constexpr to allow constant initialization.

added chips from Atlas Scientific and WeMakeThings

Atlas EZO pH, EC, ORP, DO, RTD and WeMakeThings open source i2c soil
moisture sensor.

Add PCF8574

typo