Most of the steps you did were fine, so just a few remarks:

• If you configure the server to send an UTF-8 content-type header, then this usually only works for static files like the page where your form resides in. For CGI, you are supposed to provide your own content-type (as you did), and the server must not touch them. As of today, static pages can declare their own encoding in a meta element, e.g. <meta charset="utf-8">
• Adding accept-charset="UTF-8" is good for clarity (the default value is "UNKNOWN", but not strictly required since browsers are supposed to use the encoding of the containing form if there's no accept-charset attribute.

• Your CGI script does not get any information about the encoding of characters from the request. If the browser sends UTF-8, then you need to decode the parameter accordingly.
• If you write UTF-8 in the response CGI script, you ought to print "Content-type: text/plain; charset=utf-8\n\n": In HTTP, the default charset is ISO-8859-1. Browsers can't infer the encoding if there's nothing in the Content-type header, so they'll display your two-byte é as two bytes, namely é.
• If you get a "wide character" warning (if you don't, then you forgot to decode), you might also have omitted to declare the encoding of your output stream. Perl also uses a single-byte encoding per default. To print Unicode characters as UTF-8, you need to do something like binmode STDOUT, ":encoding(UTF-8)";

About diagnostics: Windows Codepage 1252 and ISO8859-1 are different, but quite similar, so there's no way to distinguish between those two in this case.

As of today, static pages can declare their own encoding in a meta element, e.g. <meta charset="utf-8">

This is what I mean when I said I declared the charset in the HTML

If you write UTF-8 in the response CGI script, you ought to print "Content-type: text/plain; charset=utf-8\n\n":

The server had already added the UTF-8 content type in the HTTP headers

binmode STDOUT, ":encoding(UTF-8)";

I had done this but forgot to include it in the code snippet. It still didn't work.

The server had already added the UTF-8 content type in the HTTP headers

Could you please elaborate which server you are using, and how you configure it to modify the content type of a CGI script? You could also verify in your browser which encoding it uses for your text/plain response.

I also had a look at the source of CGI::Simple and found out:

• The module will decode parameters for you only if you set the global variable $CGI::Simple::PARAM_UTF8 to a true value. That's sort of difficult to guess, since it isn't documented. Of course, you can decode yourself, but it looks like you didn't.
• The module will add ; charset=utf-8 to the content type header only if you print it as print $q->header(-type => 'text/plain');, but not if you just print "Content-type: text/plain\n\n";.
So, the following just works for me:

use strict;
use warnings;
use CGI::Simple;
$CGI::Simple::PARAM_UTF8 = 1;

my $q = CGI::Simple->new();
$q->charset('utf-8');

binmode STDOUT,':encoding(UTF-8)';
print $q->header(-type => 'text/plain');
print $q->param('text'),"\n";
[download]

binmode STDOUT, ":utf8";
[download]

I had done this but forgot to include it in the code snippet. It still didn't work, unfortunately.

Sorry, but ya ain't got nothing detailed enough to be reproducible :|

and PEBKAC ;)

 
use CGI::Simple;
my $q = CGI::Simple->new();
print "Content-type: text/plain\n\n";
print $q->param('text');
[download]

Try this

use CGI -utf8; 
my $q = CGI->new; 
print $q->header(qw/ -charset UTF-8 / ),
  $q->start_html,
  $q->param('unicodes');
[download]