Make it less likely to stumble upon file name replacement

Of course that's not the right way to fix it, but it is good enough for now.
Ideally EvalbotExecuter should return the filename that was used, but current
fix is probably good enough (even though it feels weird).

evalbot.pl

@@ -298,7 +298,7 @@ package Evalbot;
         my $null    = "\N{SYMBOL FOR NULL}";
         $response =~ s/\n/$newline/g;
         $response =~ s/\x00/$null/g;
-        $response =~ s|/tmp/\w{10}|/tmp/tmpfile|g;
+        $response =~ s|/tmp/evalbot-file-\w{10}|/tmp/tmpfile|g;
         $response = IRC::FromANSI::Tiny::convert($response);
 
         my $format_len = length(encode_utf8(sprintf $format_res, $prefix, ''));

lib/EvalbotExecuter.pm

@@ -119,7 +119,7 @@ sub _fork_and_eval {
     my ($program, $executer, $ename) = @_;
 
 # the forked process should write its output to this tempfile:
-    my ($fh, $filename) = tempfile();
+    my ($fh, $filename) = tempfile('evalbot-file-XXXXXXXXXX', TMPDIR => 1);
     chmod 0644, $filename;
 
     my $fork_val = fork;
@@ -195,7 +195,7 @@ sub _auto_execute {
         $program = $executer->{program_munger}->($program);
     }
     my $cmd = $executer->{cmd_line} or confess "No command line given\n";
-    my ($prog_fh, $program_file_name) = tempfile();
+    my ($prog_fh, $program_file_name) = tempfile('evalbot-file-XXXXXXXXXX', TMPDIR => 1);
     binmode $prog_fh, ':encoding(UTF-8)';
     print $prog_fh $program;
     close $prog_fh;