Define pmapreduce as a parallel to mapreduce, and pmapbatch as a parallel to pmap

[jishnub]

This PR makes several breaking changes, but the result is perhaps more consistent now with Base and Distributed.

Now pmapreduce parallels mapreduce, so moving from a serial implementation to a parallel one can be achieved simply through a change in the name of the function.

julia> nworkers()
2

julia> @time mapreduce(x -> (sleep(0.5); x^2), +, 1:10)
  5.084111 seconds (55.17 k allocations: 3.309 MiB, 1.46% compilation time)
385

julia> @time pmapreduce(x -> (sleep(0.5); x^2), +, 1:10)
  2.882450 seconds (1.49 k allocations: 76.635 KiB)
385

The function pmapbatch now parallels pmap, except the values are necessarily passed in contiguous batches to the mapped function.

julia> pmap(identity, 1:5)
5-element Vector{Int64}:
 1
 2
 3
 4
 5

julia> pmapbatch(identity, 1:5)
5-element Vector{Int64}:
 1
 2
 3
 4
 5

Also the commutative functions are removed. An operator may be declared to be commutative through the Commutative tag.

The reduction functions are cleaned up. Several in-place reduction operators are introduced for convenience.

[codecov]

Codecov Report

Merging #9 (8ab527e) into master (c0911f8) will increase coverage by 0.16%.
The diff coverage is 99.81%.

@@            Coverage Diff             @@
##           master       #9      +/-   ##
==========================================
+ Coverage   99.71%   99.88%   +0.16%     
==========================================
  Files           8        6       -2     
  Lines        1060      846     -214     
==========================================
- Hits         1057      845     -212     
+ Misses          3        1       -2     

Impacted Files	Coverage Δ
src/ParallelUtilities.jl	100.00% <ø> (ø)
src/clusterquery.jl	100.00% <ø> (ø)
src/trees.jl	99.71% <99.54%> (-0.29%)	⬇️
src/mapreduce.jl	100.00% <100.00%> (+0.51%)	⬆️
src/productsplit.jl	100.00% <100.00%> (ø)
src/reductionfunctions.jl	100.00% <100.00%> (+3.22%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c0911f8...8ab527e. Read the comment docs.