micros() unchanged - minor systick_isr() diversion - one line added to systick_isr.

This saves a cycle - and checks both values to avoid _isr change of either value based on code/execution

Alternate micros() extends millis() using ARM_DWT_CYCCNT
> Also makes unused_interrupt_vector() weak to allow user override like Teensy3

Sketches that includes String objects did not compile, so simply copy the wstring.cpp from Teensy3 directory.

Likewise sketches that used stream functions like:
Serial1.readBytes would not compile as the Stream  sources were not included.

So  likewise copied over from Teensy3 directory

"noinline, noclone" has the opposite effect of beeing faster now on T4
Maye it would be best to use an empty define ? Then we can remove the section in the linker-file, too.

#define FASTRUN

QA quick and dirty import of the AVR emulation stuff for SPI registers.

So far none of SPCR is implemented
in SPSR only getting status of SPIF is setup
SPDR  reads and writes to LPSPI4...

So, it allows for some simple SPI write functions and wait for the write to complete and read the result.

WIth this Adafruit_ILI9341 graphic test runs

Gives more free DTCM. 
This is the first iteration - after USB and other things are done we should look at this, again.

I created a structure IMXRT_LPSPI_t
like what was done for UARTS.

I hopefully got all of the unused fields in properly to get the registers on the correct addresses.
Note: I did not update the other #define LPSPIn_ values but left them, likewise, I create a new define for each of the addresses.

That way we can double check I have everything properly aligned

Add some support for serialEventX calls when yield is called.

Currently keep table of which SerialX objects are Used (.begin called) and each has their own function which does the
if (Serial1.available()) serialEvent1();

Calling SerialX.end(), removes the call from the list.  Also the default weak pointer serialEventX functions call the disable function, as to only be called once as no one is actually processing them...

At some point will integrate this with EventResponder

Put in first pass at SerialX.end(),
Which should disable the Serial object, and I believe set the IO pins to be GPIO pins.

Also defined the extern void SerialEvent1() like functions in header in preperation for callbacks...

Hopefully will be able to get this such that only those objects that are used will be brought in by the system...

More support put in.

Now support for different formats of output.
So far I have tried 8N1, 8N2, 7E1, 7O1...

Tempsense2 address correction and added analogreadaveraging code.

Fleshing out some of the missing pieces. - Need to test
copy NVIC execution priority from T3
Set priority per Serial object

RTS pin
CTS pin - only Serial3 supports
setRX, setTX - do nothing none of these have two pins defined.

Maybe temporary - but added from previous code ability to add second buffer for RX and TX

Put in code to read all of the current stuff in Receive queue on Interrupt..

..use NVIC_NUM_INTERRUPTS instead 
i know, very minor documentational issue :)

Used macros from OneWIre to fast set/clear the optional Transmitter Enable support.

Also fixed bug in ISR where was not properly looking for TC.  This will help make SerialX->flush() to work

Wrong RX_Select value

This pass has some more support for using the Hardware Serial objects.

Including setting an ISR for doing both reads and writes,  sets up the FIFO,

Sitll lots more work to go.

Things like: fifo - need to better understand/set watermarks, also RX need to read in more than one byte if more available.

Also support fo r other features, like formats, 9 bit, ...

BUt at least you can do simple reads/writes

shifting both sides by 16 bit gives pretty good accuracy with fast code:
        movw    r3, #9155
        ldr     r2, .L2
        mul     r0, r3, r0
        lsrs    r0, r0, #1
        umull   r3, r0, r2, r0
        lsrs    r0, r0, #8
.L2:
        .word   144122641

with 984 MHz it allows nsec to be max ~143000  max which is plenty

This makes sense now, and would be useful for shorter wait-times. Could be used for PulseIn etc.  where we need defined very short delays.