Enable parallel xz compression out of the box

[bobrik]

There's no difference on smaller binaries, since they don't get spread across multiple lzma blocks to have anything to parallelize, but for bigger binaries with many dependencies there's a noticeable win.

Below are the numbers from embedding 100MB of /dev/urandom into cargo-deb itself and running it on my Macbook Air with M1 processor.

• Fast mode

real 0m3.597s
user 0m24.262s
sys  0m0.265s

• Regular mode:

real 0m9.833s
user 0m37.334s
sys  0m0.529s

[kornelski]

Nice! Thanks.