Because data is now in the 'BaseLearner'

This makes the code work with any number of neighbors. Now the
new triangle loss even works with nn_neigbors=0. I also added
that we pass the neighbors in all loss_per_interval functions.

save execution time on futures inside runners

See merge request !130

Previously we were wrapping the learner function and doing the
timing there. This makes things more complicated in the case that
we are learning an 'async def' function.

Test that runners run at all (!), and that an async function can be
used with the AsyncExecutor.

Make BaseRunner an abstract base class

Closes #107

See merge request !111

Because it breaks the examples in the docs

exponentially decay message frequency in live_info

See merge request !108

Closes #118

The code is posed on https://github.com/python-adaptive/adaptive/issues/1

See https://stackoverflow.com/a/24946631/3447047

See http://help.zenodo.org/#versioning for more information.

and remove the code to find the complete author list from git  because
our CI job will ensure that all the contributors will be included
in that file.