DietPi-Globals | G_AG*: Remove potentially harmful force options + other enhancements

[MichaIng]

Status: Testing

Commit list/description:

• DietPi-Globals | G_AG*: Remove --allow-downgrades/remove-essential/unauthenticated, since these cases are mostly harmful
• DietPi-Globals | G_AG*: Use "G_DIETPI-NOTIFY 2 '\e[0m" for initial info print to still force white text in contrast to apt-get output
• DietPi-Globals | G_AG*: Do not create input string array, since "for i in $string" has exactly the same result
• DietPi-Globals | G_AG*: "-qq" includes "-y"
• DietPi-Globals | G_AG*: Reorder apt-get command and options to have consistent "apt-get [] []"
• DietPi-Globals | G_AG*: No colour reset required after apt-get output, since G_ERROR_HANDLER implies this
• DietPi-Globals | G_AG*: Add "-q(q)" to certain commands to reduce processing output a bid

Testing:

• VM Jessie
• VM Stretch
• VM Buster

@Fourdee
I know this might break automation in some cases. However I am not keen to automate such potentially harmful APT steps. I prefer to clean up/solve cases where this then fails.

• For all APT repos we add, a key should be added as well to allow verifying authenticity.
• Essential packages should be never removed of course.
• Downgrades should be never automated, if this is attempted, then something with the packages we attempt to install is wrong and we should fix.
• Only hold states we should keep overriding via G_AGI and G_AGDUG, as legacy or user hold states can break installs. We require/expect latest packages in every case.

On Jessie, no separate hold state options exist. This means that in this case hold states will break APT. But since this should without manual user action never happen and the remaining Jessie systems is less 1%, I think this is acceptable now.