this is a saner default for large answers. instead of waiting for
probably lost/thrown away fragmented packet, do the query over TCP
instead

this fixes a problem when a module was removed, but pending
queries referenced it, causing a crash. usually when the
server was busy and a module was unloaded.
as we don’t need to copy layers at all, they’re just iterated
from the array of modules using a macro

no need to scramble queries satisfied from cache

the query flags were cleared too early, and the rec never
retried if the NS had ipv6 addresses, but all were bad

this fixes an issue when nameserver responds with AA=0 and authority
of a CNAME target (which is in current bailiwick)

DNS 0x20 https://tools.ietf.org/html/draft-vixie-dnsext-dns0x20-00
is a way to add more randomness into queries to make spoofing tougher
this implementation provides up to 32 bits of randomness to QNAME,
which is more than enough for most names (it is possible to add a
maximum of 1 bit of entropy per alphanumeric character, so it's not very
efficient with shorter names)

fixes #27