Fix missing library issue on i386/armhf

Multi-arch support and more

reduce delay :  30 seconds probably okay

File renamed outside, but not updated inside :(

even for the slowest raspberry pi on SD card ...
60 second is too long to wait for fast 64 bit intel guys who used to have mongo up-and-running in 5 or 10 seconds

Result - mongo never becomes primary on install

Removed stop scripts (no longer needed)
A bit of housekeeping

This allows post-stop to only sleep when the service exits uncleanly (like when waiting for mongo)

Added post-stop script for service restart delay

Updating my fork

some prefers this , most don't mind

restore of mongodb now implemented, thanks to @elespike !

Backwards compatibility for indices

Added --noIndexRestore so previous backups with different index versions may be restored
Added a repairDatabase call to rebuild indices and compress the database, which hopefully won't take long for large DBs and will make them run smoother
This should only be run on a master/primary DB, which is the case within the snap

rcbackup: enforce sudo, readable date

rcrestore: enforce sudo, fix 'bad system call' bug

- More precise directory selection just in case directory names are different and unexpected

- Better error checking
- Code quality

- Back to using $SNAP_COMMON (so sudo can be enforced)

Create rcrestore command

`rs.initiate()` causes default to sometimes, under some runtime environments, to select the wrong hostname.

they need the .node_version.txt  file to exist

Added some niceties like timestamps and logs