OK, I shortened it, hope this is more acceptable. In the process I discovered, as I suspected, that it has something to do with the regex order, but exactly what is still pretty opaque to me.

#!/usr/bin/perl

my @words = ('aa', 'as', 'es', 'is', 'os', 'sh', 'si', 'so');
print "WORDS: @words\n";

print "----------------------------\n";

my $regex = ".?a?";
print "Matches for regex pattern : $regex\n\n";

foreach (@words) {
    if(join('', sort split //, $_) =~ /^$regex$/) {
        print "$_\n";
    }
}

print "----------------------------\n";

my $regex = "a?.?";
print "Matches for regex pattern : $regex\n\n";

foreach (@words) {
    if(join('', sort split //, $_) =~ /^$regex$/) {
        print "$_\n";
    }
}

print "----------------------------\n";

$regex = "s?.?";
print "Matches for regex pattern : $regex\n\n";

foreach (@words) {
    if(join('', sort split //, $_) =~ /^$regex$/) {
        print "$_\n";
    }
}


print "----------------------------\n";

$regex = ".?s?";
print "Matches for regex pattern : $regex\n\n";


foreach (@words) {
    if(join('', sort split //, $_) =~ /^$regex$/) {
        print "$_\n";
    }
}


print "----------------------------\n";
[download]

OUTPUT

WORDS: aa as es is os sh si so
----------------------------
Matches for regex pattern : .?a?

aa
----------------------------
Matches for regex pattern : a?.?

aa
as
----------------------------
Matches for regex pattern : s?.?

----------------------------
Matches for regex pattern : .?s?

as
es
is
os
sh
si
so
----------------------------
[download]

Regards.