rpc: refactor subscriptions and filters

TravisCI: add Go 1.7 targets, bump PPA to 1.7 too

core: ensure the canonical block is written before the canonical hash is set

eth/downloader: fewer headers and futures too in ancestor lookup

Fixes #2872

rpc: client bug fixes

eth/downloader: abort sync if master drops (timeout prev)

eth, eth/downloader: don't forward the DAO challenge header

eth, eth/downloader: better remote head tracking

Makefile, build: move cross compilation into ci.go

core/vm: hide ecrecover error message

Fixes #2825

It's inconsistent not to pass it and most callers will
work with contexts anyway.

I initially made the client block if the 100-element buffer was
exceeded. It turns out that this is inconvenient for simple uses of the
client which subscribe and perform calls on the same goroutine, e.g.

    client, _ := rpc.Dial(...)
    ch := make(chan int) // note: no buffer
    sub, _ := client.EthSubscribe(ch, "something")
    for event := range ch {
        client.Call(...)
    }

This innocent looking code will lock up if the server suddenly decides
to send 2000 notifications. In this case, the client's main loop won't
accept the call because it is trying to deliver a notification to ch.

The issue is kind of hard to explain in the docs and few people will
actually read them. Buffering is the simple option and works with close
to no overhead for subscribers that always listen.

Godeps: update github.com/rjeczalik/notify to f627deca7a51

Fixes #2829

node, p2p, internal: Add ability to remove peers via admin interface

Fix README typo

"fine graned" -> "fine-grained"

containers/docker: use shallow clone of single branch

eth/downloader: fix the stall checks/drops during sync

rpc: add new client, use it everywhere

eth, eth/downloader, eth/fetcher: delete eth/61 sync code

The new client implementation supports concurrent requests,
subscriptions and replaces the various ad hoc RPC clients
throughout go-ethereum.

If a batch request contained an invalid method, the server would reply
with a non-batch error response. Fix this by tracking an error for each
batch element.

The server delayed closing of connections for 3s when stopping. This was
supposed to allow for slow handlers, but it didn't really work. When
geth quits, it will just exit immediately after quitting the server.

Removing the timer makes testing easier because all connections will be
closed after Stop returns.

eth/api: rename signAndSendTransaction to sendTransaction