Do not abort in case of invalid user input but retry

[jsmeix]

• Type: Enhancement

• Impact: Low

• Reference to related issue (URL):
  Do not abort 'rear recover' in case of invalid user input - let the user retry if possible #2253

• How was this pull request tested?
  I tested the plain code by running it as a separated script
  and for me the input seems to work o.k.
  but I cannot actually test it with TSM or NBU
  because I have neither of them.

• Brief description of the changes in this pull request:

Let the user enter date and time again and again until the input is valid
or the user pressed only ENTER to restore the most recent available backup.

Replaced 'LogPrint' with 'UserOutput' because
the former shows messsages only in verbose mode
while the latter shows messages always to the user
(this does not matter in practice during 'rear recover'
but 'UserOutput' is the ceaner code for user output
and I noticed that in my separated test script).

[jsmeix]

@schabrolles
perhaps -if time permits - you could even test it with TSM?

@gdha
I would appreciate it if you could have a look at the plain code,
perhaps you notice something suspicious?

[jsmeix]

There is also

verify/GALAXY10/default/550_request_point_in_time_restore_parameters.sh

which could benefit from such an enhancement.

The following scripts look also suspicious:

# find usr/share/rear -name '*.sh' | xargs grep -l WAIT_SECS

usr/share/rear/restore/DP/default/460_press_y_to_continue.sh
usr/share/rear/restore/DUPLICITY/default/400_restore_duplicity.sh
usr/share/rear/verify/default/040_validate_variables.sh
usr/share/rear/verify/DP/default/500_select_dp_restore.sh
usr/share/rear/verify/DP/default/450_request_gui_restore.sh
usr/share/rear/verify/NBU/default/380_request_client_destination.sh
usr/share/rear/verify/TSM/default/400_verify_tsm.sh

For next week (if time permits)...

[jsmeix]

When there are no objections I would like to "just merged" it
so that things could be tested by users "out there"
as long as ReaR 2.6 is still under development.

[jsmeix]

I think this one is sufficiently complete and consistent now
so that - if there are no objections - I would merge it today afternoon.

I kept WAIT_SECS as separated user config variable for now
but I think in the future WAIT_SECS should be removed
and only USER_INPUT_TIMEOUT should be kept and
the remaining direct read calls should be replaced
by UserInput in a future ReaR 3.0 version, cf.
#1390
and
#1705

[jsmeix]

I "just merged" it so that things can be tested right now
as long as ReaR 2.6 is still under development.

Please report if there are regressions so that I can fix them,
ideally each one as a new and separated GitHub issue.

[jsmeix]

With #2257 merged plus
08f5956
I think this issue is solved as far as possible for me for now
and hopefully even without regressions.

I would appreciate it if ReaR users in particular those who
use BACKUP=TSM or BACKUP=NBU or BACKUP=GALAXY10
could test that changes as long as ReaR 2.6 is still under development
and report if there are regressions so that I can fix them,
preferably each one as a new and separated GitHub issue.