Use multithreaded load / save per entry

[haampie]

Loading 10GB of data on a laptop with 4 cores goes from

julia> example()
 43.730908 seconds (11.57 k allocations: 9.866 GiB, 2.12% gc time)

to

julia> example()
 14.883994 seconds (11.60 k allocations: 9.866 GiB, 1.79% gc time)

Saving is probably even more interesting, that took 6mins, I can check if it goes to 1.5 minutes with this PR

[oxinabox]

Nice. still some issues.
But you got the locks in the right places which is important

Can we do this in a way that maintains compatibility with Julia 1.0?
Does Compat.jl have support for making @spawn into @async or some such?

[haampie]

I'll take a look at Julia 1.0 compat a bit later

[codecov]

Codecov Report

Merging #51 into master will increase coverage by 0.15%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master      #51      +/-   ##
==========================================
+ Coverage   96.00%   96.15%   +0.15%     
==========================================
  Files           6        6              
  Lines         150      156       +6     
==========================================
+ Hits          144      150       +6     
  Misses          6        6              

---

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 9531c5f...b6cf74c. Read the comment docs.

[oxinabox]

LGTM
can you bump the minor version so we can tag a release?

[oxinabox]

Thanks!