add flushdb option for repl-diskless-load

[kronwerk]

once upon a time it was an initial way to make diskless replication - without any checks.
but it was frown away because it's dangerous (and it is still dangerous, of course).
but in some situations for some users, that think they know what they're doing, it's ok.

let's pls allow this for those. thx

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.61%. Comparing base (9f8185f) to head (8116d29).
Report is 7 commits behind head on unstable.

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable     #909      +/-   ##
============================================
+ Coverage     70.59%   70.61%   +0.01%     
============================================
  Files           114      114              
  Lines         61670    61671       +1     
============================================
+ Hits          43538    43550      +12     
+ Misses        18132    18121      -11     

Files with missing lines	Coverage Δ
src/config.c	78.69% <ø> (ø)
src/replication.c	87.91% <100.00%> (+0.66%)	⬆️
src/server.h	100.00% <ø> (ø)

... and 16 files with indirect coverage changes

[PingXie]

I think this is a reasonable policy to add and it doesn't increase the complexity.

@valkey-io/core-team

[enjoy-binbin]

i am ok with this, i mention this option in #653 (comment)

[zuiderkwast]

It seems OK, but maybe risky that some users lose data by mistake.

but in some situations for some users, that think they know what they're doing, it's ok.

What is the use case? Why don't they want to explicitly call FLUSHALL before starting replication?

If we require them to do it explicitly, it is less likely that some user will lose data by mistake.