Unexpected CONTINUATION frame is handled separately as connection
error.

I have this error with some old gcc (4.6) release (Ubuntu 12.04 or Travis...)

Mainly make nested code block to rather flat style.

Previously, we use evbuffer_pullup(buf, -1) to linearize the memory
region and it may cause buffer copy.  To avoid this, we use the return
value of evbuffer_get_contiguous_space() as 2nd parameter.  According
to the libevent manual, by doing so evbuffer_pullup() will not copy or
modify any data in evbuffer.

Setting write timeout to filter bufferevent does not work as intended.
It timeouts even when there is no data to write.

Previously we have uint16_t as state member variable in
nghttp2_huff_decode structure to express -1 as failure.  This is
because we have 256 valid states.  However, we can express failed
state using flags member variable and make state uint8_t.  This commit
does this and as a result the size of decoding table is reduced.

Previously we use 2 separate buffer for each name and value.  The
problem is we would waste buffer space for name because it is usually
small.  Also tuning buffer size for each buffer separately is not
elegant and current HTTP server practice is that one buffer for 1
name/value pair.  This commit unifies 2 buffers into 1.

Use gcc-4.8 (don't build with gcc-4.6)
Update libstdc++-4.8 to fix Clang build

Don't install cython (Never build...)

It seems that specifyig '*' to node parameter in getaddrinfo() is
treated as specifying NULL, but it is not documented.  So rather than
relying on this feature, we explicitly treat '*' as "wildcard" address
and specify NULL to node parameter in getaddrinfo().

Now '*,3000' is a default value of --frontend option.  Specyfing '*'
binds all addresses including both IPv4 and IPv6.

We are currently disabled ALPN in nghttp2.org