Replace mruset setAddrKnown with CRollingBloomFilter addrKnown

Use a probabilistic bloom filter to keep track of which addresses
we think we have given our peers, instead of a list.

This uses much less memory, at the cost of sometimes failing to
relay an address to a peer-- worst case if the bloom filter happens
to be as full as it gets, 1-in-1,000.

Measured memory usage of a full mruset setAddrKnown: 650Kbytes
Constant memory usage of CRollingBloomFilter addrKnown: 37Kbytes.

This will also help heap fragmentation, because the 37K of storage
is allocated when a CNode is created (when a connection to a peer
is established) and then there is no per-item-remembered memory
allocation.

I plan on testing by restarting a full node with an empty peers.dat,
running a while with -debug=addrman and -debug=net, and making sure
that the 'addr' message traffic out is reasonable.
(suggestions for better tests welcome)

src/net.cpp

@@ -1905,7 +1905,7 @@ bool CAddrDB::Read(CAddrMan& addr)
 unsigned int ReceiveFloodSize() { return 1000*GetArg("-maxreceivebuffer", 5*1000); }
 unsigned int SendBufferSize() { return 1000*GetArg("-maxsendbuffer", 1*1000); }
 
-CNode::CNode(SOCKET hSocketIn, CAddress addrIn, std::string addrNameIn, bool fInboundIn) : ssSend(SER_NETWORK, INIT_PROTO_VERSION), setAddrKnown(5000)
+CNode::CNode(SOCKET hSocketIn, CAddress addrIn, std::string addrNameIn, bool fInboundIn) : ssSend(SER_NETWORK, INIT_PROTO_VERSION), addrKnown(5000, 0.001, insecure_rand())
 {
     nServices = 0;
     hSocket = hSocketIn;