Something has shifted.   The previous timing delays on the Max7219 debug displays is too tight without this correction.
I suspect something has been optimized and roughly 50ns of needed setup and hold time has disappeared.
This corrects the issue and the display results are clean again.

Some of the Rotations did not do the right thing with setting and clearing pixels.
I think it is correct now.   But if not...  It is much closer to being correct.

Address #11463

* Add Multi Unit Support for Max7219 to bugfix_2.0.0

* Apply multi-MAX7219 to example configs

* Tweak some spacing, macros

Disabling an ISR on ARM has 3 instructions of latency. A Memory barrier is REQUIRED to ensure proper and predictable disabling. Memory barriers are expensive, so avoid disabling if already disabled (See https://mcuoneclipse.com/2015/10/16/nvic-disabling-interrupts-on-arm-cortex-m-and-the-need-for-a-memory-barrier-instruction/)

Use macros that explicitly avoid double-evaluation and can be used for any datatype, replacing `min`, `max`, `abs`, `fabs`, `labs`, and `FABS`.

Co-Authored-By: ejtagle <ejtagle@hotmail.com>

Co-Authored-By: ejtagle <ejtagle@hotmail.com>

on the STM32 platform

A number of regressions were patched also. The UBL G29 P2 and P4 Press and Hold had stopped working. It is very possible this is broken in the bugfix_v1.1.x branch also.

The main purpose of the Pull Request is to get the 3-Point mesh tilting to use the LSF algorithm just like the grid based mesh tilt. This simplifies the logic and reduces the code size some what. But the real reason to do it is the 3-Point case can be solved exactly. And by feeding these numbers into the LSF algorithm it provides a way to check all that code for 'correctness'.