Bugfix

[xanrag]

Quite a few changes you might want to cherrypick something from, top to bottom:

• change retries to 1 and set a timeout of 10s on fetch_ip
• rewrote the check_alive function to be much simpler, the fetch_ip check was not viable at 20+ proxies. It took too long.
• updated ip_list not to return ip:s slated for destruction
• added option to restart AWS proxies, much faster than destroy/create and fetches a new IP. Not supported for DO.
• opened up port 22 on proxies for debugging, future enhancement is to only allow web control. (Use the EC2_INSTANCE_CONNECT filter for the service parameter to get the IP address ranges in the EC2 Instance Connect subset. )
• enhanced status messages a bit in the check_alive for AWS
• moved check_delete/stop to before provision so it removes and then immediately provisions a new one instead of waiting 20s for next tick
• changed the proxy software to tinyproxy directly on the image instead of using docker. Much faster deployment and less cpu intensive so should work better with t2.nano
• updated the settings checks to compare true/false as a string since it seems to be what it is getting, earlier a value of False in the config would read as true.
• updated environ get to match the doc (ie SCALING instead of SCALE)
• added botocore to requirements.txt, seemed to be missing.

[claffin]

Thanks for the pull request, I've reviewed the changes, tested and happy to merge them. Good idea to switch to tinyproxy.

Thanks for your contribution! Sorry for the late review.