I think the following will give you a very good spread. I concocted it from the top of my head, so I don't garantee results.

use strict;
use warnings;


open(my $fh, '<', '...')
or die("Unable to open input file: $!\n");


# Group domains into an AoA.
# Very efficient since the input fileis already sorted by domain.

my @grouped;

my $prev_domain = 'invalid domain';

while (<$fh>) {
   chomp;

   my $addr   = $_;
   my $domain = ...;

   if ($domain ne $prev_domain) {
      $prev_domain = $domain;
      unshift(@grouped, []);
   }

   push(@{$grouped[0]}, $addr);
}


# Sort groups by group size.
# Gives something like:
# @sorted = (
#    [ qw( g1             ) ],
#    [ qw( f1             ) ],
#    [ qw( e1             ) ],
#    [ qw( d1 d2          ) ],
#    [ qw( c1 c2 c3       ) ],
#    [ qw( b1 b2 b3       ) ],
#    [ qw( a1 a2 a3 a4 a5 ) ],
# );

my @sorted = sort { @$a <=> @$b } @grouped;


my @ordered;

my $prev_last = 0;
my $break     = 0;
my $offset    = 0;

for (;;) {
   my $p = shift(@sorted);

   last if not defined $p;

   my $last = $#$p;
   if ($last != $prev_last) {
      $prev_last = $last;
      $break     = @ordered / $last;
      $offset    = 0;
   }

   foreach my $i (reverse 0..$last) {
      splice(@ordered, $i*$break+$offset, 0, $p->[$i]);
   }

   $offset++;
}

{
   local $, = ", ";
   local $\ = "\n";
   print(@ordered);
}
[download]