This is why I wanted to sleep on the code I wrote while falling asleep
rather than immediately submitting a pull request.

This seems to defeat the purpose of previous_millis_cmd, preventing the
time elapsed from previous_millis_cmd from ever reaching
max_inactive_time or stepper_inactive_time while the heat was on.

Fixes a bug in the Extruder Runout Prevention feature that caused the
extruder to move back to “current_position” after a move if it was
activated while the move was in progress. For long home moves (which
are longer than the dimensions of the machine’s working area), this
would cause the machine to crash into the far end of its travel after
homing. This usually occurred on the Z axis, which could result in
damage to the machine if you don’t hit the reset button in time.

The code for the LSQ solver (qr_solve) is copyrighted by John Burkardt and released under LGPL here:
http://people.sc.fsu.edu/~%20jburkardt/c_src/qr_solve/qr_solve.html
(see qr_solve.cpp for further copyright information)

Removed the calculation of the inverse matrix since the rotation matrix is orthogonal, therefore inverted == transposed.
Much simpler and mathematically robust.

In some cases the Bed Heater FET heats up more then stepper drivers, so
this change add the bed monitoring to the controller fan. As soon as the
bed heater is turned on, the controller fan will run as well.

Recommended for those who are using the Z Probe for Z Homing (as
Z-Endstop)

This feature has two changes:

1) Allow user to choose where the Z Probe will touch the bed when homing
all axis together (G28) by setting below defines:

Z_SAFE_HOMING_X_POINT
Z_SAFE_HOMING_Y_POINT

2) Prevents the user to perform Z Axis Homing when the Z Probe is
outsite bed.

Implemented M226 as described here:
http://reprap.org/wiki/G-code#M226:_Gcode_Initiated_Pause

Waits for pin to be become either HIGH, LOW or the inverse of what it
was before. Allows printing to pause until user interaction

start another print with suicide feature after a print without
suicide...

Allows the user to select wheter or not the PSU should be turned on or
kept in standby when marlin boots

since it can occure in comments.

by overloading M32 it is now possible to execute gcode files from other gcode files, with a fixed recursion level.
This can be used e.g. for having a real start.g and end.g somewhere on the sd card, which are then called from the normal print file.
Another usecase would be to have macro-files for nozzle-change and layerchange.
I have not tested the speedwise performance. The testing was done with pronterface.

syntax:
normal call from sd card will open the new file and continue executing there.
M32 !/path/filename#
this however will call the new file and return to the caller file.
M32 P !/path/filename#
with the optional "S<position>" the  file starting position can be set.
this is for continuing prints from a previous location.

if a '#' is read now the buffer will be emptied before reading ahead.
This is so one can execute files from within gcode files, without messing the buffer with preread characters from the caller file.
# can not occure in sd files imho, because it should only occure within checksums in ther serial communication.
Yes, thats a lame argument. If you have a better idea please tell me. It has to be a character that one can type
on a keyboard manually.

Hopefully fixed viky button handling without braking other boards

- Added "Z_RAISE_BEFORE_HOMING" for raising Z the defined distance
before homing. This is useful to avoid Z-Probe collision when hotend is
near bed.

- Fixed the issue of Z not going bellow Z_PROBE_OFFSET when
"min_software_endstops" is true.
Now the Z_PROBE_OFFSET is not set in Z_MIN_POS, it is added after
homing.

If "SHOW_TEMP_ADC_VALUES" is defined in Configuration_adv.h, the M105
command will present, after tradicional temperatures, the ADC value read
from temp sensors. This is great for adjusting thermistor tables with
thermocouple.

From Pronterface you can see the ADC value and compare with a
thermocouple reading.. then you just need to create your own thermistor
table.

Since this merge doesnt change the original information, it doesnt mess
with PC software parsing (tested under Pronterface and Repetier-Host).

Check the Readme for instruction how to enable and configure the feature

As well a couple of minor changes to handling of extruder offset for
dual x-carriage mode.

Also added a couple of missed merge lines in homeaxis() from previous
commit.

add more delta defaults
on delta make second home even slower

Dual x-carriage designs offer some substantial improvements for dual
extruder printing.

Untested

M109 S180 waits only when heating.
M109 R180 also waits when cooling.

Diffed and merged, preserving my updates