Corrected timeout increment between requests

[homedirectory]

According to the RFC:

To determine the external IP address or request a port mapping, a NAT-PMP client sends its request packet to port 5351 of its configured gateway address, and waits 250ms for a response. If no NAT-PMP response is received from the gateway after 250ms, the client retransmits its request and waits 500ms. The client SHOULD repeat this process with the interval between attempts doubling each time. If, after sending its 9th attempt (and then waiting for 64 seconds), ...

Current implementation, however, always increments the timeout by 0.250 seconds, instead of doubling it. So after the 9th attempt the timeout would only be 2.25 seconds instead of desired 64.

[yimingliu]

Hi!

Thank you for all the contributions! Really appreciate your work on a library that I have not worked on in many years.

Some asks:

• The original patch looks safe to integrate. With this package lacking a real test suite, however, it would be appreciated if you showed run results (perhaps with the test client) that the changes did not break anything.
• Subsequent patches are proposing substantial changes. Can we separate these out as individual, self-contained PRs, for clarity of review?
• Certain changes are quite aggressive for a library. For example,
  • the removal of symbols like NATPMP_PROTOCOL_UDP and such from the module namespace.
  • forward_port changing from member method to staticmethod
  • Downstream code could be importing these symbols, or relying on their presence in the module space. Are you able to do what you want, without creating breaking changes?
• In setup.py, this library claims to support Python 2.7 only. Are we proposing to upgrade to Python 3 support only? If so, configuration changes should be made to that file as well so that it no longer makes that claim. Otherwise, careful testing should be done on both 2.7 and 3.x.

Again, definitely welcome contributions, but there is some concern here for making aggressive changes beyond the scope of the original PR. Let's take it easy and one step at a time?

Thank you!

[homedirectory]

@yimingliu Sorry, I forgot that I was still comitting to this PR's branch, so only the first commit is relevant. Otherwise, I agree with you on all points. I will recreate the PR and include run results.