Increase dnscrypt-proxy max clients to 300
Switch to IPv4 internally

Trying to narrow down an issue where the server sometimes is really
slow to response.
Raise soacceptqueue and enable accf_http/accf_data for Nginx
Apply hardening playbook
Move SSH port to 4022
Use fifo to pass Ansible password instead of stdin
Massive cleanup and move to
Fix nginx configuration mis-redirecting letsencrypt requests.
Add Japanese-site specific blocklist.
Properly setup logger to discard request logs.

Due to misconfiguration, all requests prior to Sep 4, 2018 19:00 UTC
were logged by doh-proxy, which prints to STDOUT, and s6-svscan-log
capture those to /tmp/uncaught-logs.

In this commit I've proceed to properly configure s6 logger to discard
request logs from doh-proxy, as well as deleting all logs in the
server. I deeply apologize for not properly auditing this. :'(
Enable IPv6 and add cloudflare-ipv6 as DoH upstream.
Expose unbound directly in additional to DoH and some optimizations.
Force dnscrypt-proxy to use TCP and disable cache.

Another attempt at fixing SERVFAIL.
Disable negative caching in unbound.

Another attempt to fix random NXDOMAIN.
Attempt to fix random NXDOMAIN.
Use absolute path for emptyenv.
Another attempt to fix periodic.
Cleanups and attempt to fix periodic.