get rid of twi.h

move to board files

Setup FolgerTech i3-2020 Configuration files as a reference platform for
32-Bit work.

Also fix MAX7219 debug lights on 32-bit platforms.

Same as #7728 but for 2.0.x,
Lot of cleanup and remove references in whole code to other "LED files" than leds.h. Now will be much easier to add next drivers/libraries. e.g. FastLED. But bad news, currently FastLED is suporting only RGB devices (no RGBW)

Compiler error, added char c variable

Fix double character sent when TX buffer size is 0 on mega2560

* Auto-enable DISABLE_REDUCED_ACCURACY_WARNING if HOME_AFTER_DEACTIVATE enabled

If HOME_AFTER_DEACTIVATE, there is no need to warn on LCD since printer will home prior to starting print.  Saves 60 bytes PROGMEM on AVR with U8g panel.

* Reduce lines used

* Reduce lines used

Alternative to #7882. If F_CPU is greater than 1000 it can be evenly divided by 8. Over 10000, 16; over 100000, 32; over 1 million, 64; etc.

Add to controllerfan.cpp:
#include "../module/stepper_indirection.h"
#include "../module/temperature.h"

Change "extruder" to HOTEND in sanity error messages

This change initialize any data in eeprom.dat beyond the current file size to FF.
That way if eeprom.dat is deleted and created again, it doesn't take the old values or random ones, but rather starts with FF in all positions as a real brand new or erased eeprom.dat
Currently if you delete eeprom.dat and restart the board, the new file is created in the same sector with the same content, since FAT does not actually delete the data, just marks the sector as free. I tested by deleting the file, and then rebooting the board, and checking the file content.
The change can be tested in the same way, deleting, rebooting the board, and then the new content should be all FF.

If an eeprom file already exist with data on it, but smaller than E2END, it will be padded with FF on first access, so it will not have random or old content appended.

Derived from #7873

Adapted from #7865

Based on #7862, #7846 by @rmoravcik

As pointed out in #7692 by @totalitarian

* test only

test

* Fixed RAMPS FD V1 pins.h file

* Fix typo

When the buffer is empty index_write == index_read, but they needed constrained by buffer_mask

* fixed some include paths

* LPC1768: Fix Serial API

Add missing serial methods used if TX_BUFFER_SIZE is set
Change return value of HalSerial:read to match Arduino API

* LPC1768: add filters to ADC

This is to try and compensate for hardware issue and oversensitivity to noise

* LPC1768: remove the polling section of delayMicroseconds

* LPC1768: lock usb mass storage device while device accesses it.

Currently only applicable to persistent store,
The device always has priority and will unmount the sd card from the host, Windows then tries to automount again so it can look like the explorer window freezes. Linux Mint, by default, just closes the Nemo window.

* Add timeout to make sure if Serial never connects that Marlin still boots

* Remove unneeded ifdef CPU_32_BIT

In general the need for ifdef CPU_32_BIT blocks means that something is missing from the HAL API or a Platform, in this case HAL_TICKS_PER_US was missing from the AVR Platform

* LPC1768: relocate RE-ARM debug_extra_script.py

From @ejtagle, originally #7459

Derived from #7811 and #7779