Allow bitarray to handle shifts by a negative value.

[StefanKarpinski]

Rebased and fixed conflicts here: #18154

[tkelman]

@nanosoldier runbenchmarks(ALL, vs = ":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @jrevels

[tkelman]

there is the doc issue @rfourquet mentioned, opened as #19767 if you don't want to get it now

[StefanKarpinski]

Any clue what the deal with the failed AppVeyor run is? Seems random and unrelated.

[StefanKarpinski]

There are a handful of apparent regressions (and one progression), but most of them seem to be unrelated and just noise, with the exception of the nullable arrays regression, which could be real and is fairly significant – 50% slow down.

[tkelman]

appveyor failure looks like #16555

[StefanKarpinski]

How about I merge this and open a performance issue for the one regression?