There's no way to do this in general from a shell. What you have to do is run the terminal program itself, or some launcher program that does so for you. And the way to do that is different for each terminal program.
In some cases, os.startfile will do what you want, but this isn't going to be universal.
Also, note in general, you're going to actually need an absolute path to your script, because the new terminal window will be running a new shell and therefore won't necessarily have your same working directory. But I'll ignore that for the examples.
With Windows cmd, the easiest way to do it is the start shell command. If the thing you start is any command-line program, including python, it will get a new cmd window. So, something like:
subprocess.call('start /wait python bb.py', shell=True)
OS X has a similar command, open. And it's a real program rather than a shell command, so you don't need shell=True. However, running a command-line program or script with open doesn't generally open a new terminal window. In fact, the whole point of it is to allow you to run programs as if they were being double-clicked in Finder, which never runs something in the terminal unless it's a .command file.
So, you can create a temporary .command wrapper file and open that; something like this (untested):
with tempfile.NamedTemporaryFile(suffix='.command') as f:
f.write('#!/bin/sh\npython bb.py\n')
subprocess.call(['open', '-W', f.name])
Alternatively, you can explicitly tell open to use Terminal.app, something like this:
subprocess.call(['open', '-W', '-a', 'Terminal.app', 'python', '--args', 'bb.py'])
Or you can script Terminal.app via AppleEvents. For example:
appscript.app('Terminal').do_script('python bb.py')
The "do script" event opens a new window and runs its argument as a command. If you want more detailed control, open the scripting dictionary in AppleScript Editor and see all the fun stuff you can do.
On Linux or other *nix systems… well, there are 65,102 different desktop environments, launchers, and terminal programs. Do you need to work on all of them?
With gnome-terminal, just running the terminal again gives you a new window, and the -x argument lets you specify an initial command, so:
subprocess.call(['gnome-terminal', '-x', 'python bb.py'])
Many older terminals try to be compatible with xterm, which does the same thing with -e, so:
subprocess.call(['xterm', '-e', 'python bb.py'])
subprocess.call(['rxvt', '-e', 'python bb.py'])
… etc.
How do you know which terminal the user is using? Good question. You could walk the like of parent processes from yourself until you find something that looks like a terminal. Or you could just assume everyone has xterm. Or you could look at how various distros configure a default terminal and search for all of them. Or…
