replication: fix io-threads possible race by moving waitForClientIO

[uriyage]

Fix race with pending writes in replica state transition

The Problem

In #60 (Dual channel replication) a new connWrite call was added before the waitForClientIO check. This created a race condition where the main thread may attempt to write to a client that could have pending writes in IO threads.

The Fix

Moved the waitForClientIO() call earlier in syncCommand, before any connWrite call. This ensures all pending IO operations are completed before attempting to write to the client.

[ranshid]

@uriyage I am fine with the fix but let's also introduce assertion or debug assertion in connWrite to catch these things.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.89%. Comparing base (789a73b) to head (bbc1829).
Report is 98 commits behind head on unstable.

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable    #1422      +/-   ##
============================================
+ Coverage     70.66%   70.89%   +0.22%     
============================================
  Files           114      119       +5     
  Lines         63150    64652    +1502     
============================================
+ Hits          44624    45832    +1208     
- Misses        18526    18820     +294     

Files with missing lines	Coverage Δ
src/replication.c	87.29% <100.00%> (+0.03%)	⬆️

... and 55 files with indirect coverage changes