GitLab

This commit extends the bitswap protocol with two additional wantlist properties:

* WANT_HAVE/HAVE: Instead of asking for a block, a node can specify that they
  want to know if any peers "have" the block.
* WANT_HAVE_NOT/HAVE_NOT: Instead of waiting for a timeout, a node can explicitly
  request to be told immediately if their peers don't currently have the given
  block.

Additionally, nodes now tell their peers how much data they have queued to send
them when sending messages. This allows peers to better distribute requests,
keeping all peers busy but not overloaded.

Changes in this PR are described in: https://github.com/ipfs/go-bitswap/issues/186

overlap3 and 2 are identical

add proper usage of go benchmark timing and environment random seed support for CI

Instance generator was previously named session generator, which created confusion with bitswap
sessions

fix #101

Add comments to all exported functions, extract the utils for creating instances in testnet.go,
moves integration tests to bitswap_test

BREAKING CHANGE: removed one constant -- rebroadcastDelay -- which I believe was unused

Limits connection bandwidth in real world benchmarks so that blocks are delayed if single peer is
overused

fix #40

make both performance benchmarks write to a tmp dir and put in the .gitignore

- moved sessions out of main bitswap package
- modified session manager to manage all sessions
- moved get functions to their own package so sessions can directly

BREAKING CHANGE: SessionsForBlock, while not used outside of Bitswap, has been removed, and was an
exported function

- add a delay generator that similates real world latencies one might encounter on the internet
- modify virtual network to accept different latencies for different
peers based on using NextWaitTime on passed delay
- modify dup_blocks_test subtestDistributeAndFetch to accept a custom
delay
- Add a real world benchmarks that simulates the kinds of problems one might
encounter bitswaping with a long lived session and a large swarm of
peers with real world latency distributions (that causes #8 not to
function well in practice)

So that CI passes, and because it's not reliable as a test, and is more a benchmark to measure
performance, convert dup_block_test.go to a benchmark, which can be run using `go test -bench .`