How would running all of the .t files using TAP::Harness from within a .t file, rather than from the main script give me any different output?

The only way I can imagine getting a different output would be to run it something like this:

set SDTOUT to a file 
for each .t file
   run .t
   parse output file
   for each line do either:
      is(1,1,testcasename) or is(1,0,testcasename)
next file
[download]

What is is that you're suggesting?

Something like this:

use strict;
use warnings;
use Test::More;
use TAP::Harness;

my $harness = TAP::Harness->new({verbosity => -9});
my $agg_results = $harness->runtests(glob "suite_foo/*.t");
my @tests = $agg_results->descriptions;
plan(tests => scalar(@tests));
for my $test (@tests) {
   my ($parser) = $agg_results->parsers($test);
   ok(! $parser->has_problems, $test);
}
[download]

Note that the new TAP::* modules have some documentation infelicities: TAP::Parser's has_problems says it complains about passed TODO tests but it doesn't (and shouldn't), and TAP::Harness verbosity < -2 (silent) isn't documented.
--
Online Fortune Cookie Search