PID tuning, disturb vs command pos?

I'm interested in learning more about various PID tuning methods, any good online resources or books would be appreciated. I've got some experience PID tuning pump control systems and with my mills previous step and direction drives but am fairly new to using the HALScope to do it with a full closed loop thru mesa hardware (7i29 drives and a 5i20 FPGA)

I saw this link thru the linuxcnc wiki:
wiki.linuxcnc.org/cgi-bin/wiki.pl?PWM_Servo_Amplifiers

which suggests disturbing the servo's position in order to see how fast it settles and what numbers need to be changed.

I'm curious what benefits or detriments this method has versus just running a gcode program that rapids an axis back and forth over a few inches, or makes a circular motion. For that matter I'm not even sure if linuxcnc will allow configuration changes while a program is running - it was just a thought I had tonight.

I'm wondering about this because I seem more readily able to make the axis I chose to tune first "ring" by jogging than by just pushing the handle one way or the other

Also, I've been editing my numbers by using axis -> machine -> configuration, is there any way to change numbers easier (like tying movement on a decimal place to the scrollwheel on my mouse or a griffin powermate or something?) or not having to constantly hit OK to enter new numbers?

Better yet, is there a way to tie config changes into HAL so I'm not switching windows constantly? This would seem like a better way to do things for those of us who don't have huge monitors or resolutions to play with on our cnc's.

The PID component is linear and symmetrical so the commanded position and
feedback position are equivalent as far as entering disturb values goes.

(edit: until you add FF terms)

I expect any difference you see is due to the different frequency components
in a feedback position disturb caused by pushing the axis (limited bandwidth)
and a commanded position disturb cause by jogging (much higher bandwidth)

Axis has a tuning helper 'calibrate' if you are not already using it. The calibrate routine
can write the current tuning values back to your hal file