Use rayon when running `bootimage test`

[acheronfail]

Fixes #21

This PR adds rayon and uses it during the bootimage test command so that all the tests are run in parallel.

There's a few caveats though:

• While the tests are building they block each other while waiting to build
• Their output sometimes mangles each other since the calls are done in parallel

Aside from those, the actual tests themselves are run in parallel, just not when they're built..

Suggestions welcome for improvements!

[phil-opp]

Very nice, thanks!

While the tests are building they block each other while waiting to build

This is a bit unfortunate. It also causes some very verbose output:

``` Blocking waiting for file lock on x86_64-blog_os's sysroot Blocking waiting for file lock on x86_64-unknown-linux-gnu's sysroot Blocking waiting for file lock on x86_64-unknown-linux-gnu's sysroot Blocking waiting for file lock on x86_64-unknown-linux-gnu's sysroot Blocking waiting for file lock on build directory Blocking waiting for file lock on build directory Finished dev [unoptimized + debuginfo] target(s) in 0.34s Building bootloader v0.3.4 Finished dev [unoptimized + debuginfo] target(s) in 0.36s Building bootloader v0.3.4 Finished dev [unoptimized + debuginfo] target(s) in 0.39s Building bootloader v0.3.4 Finished dev [unoptimized + debuginfo] target(s) in 0.05s Building bootloader v0.3.4 Blocking waiting for file lock on x86_64-bootloader's sysroot Blocking waiting for file lock on x86_64-bootloader's sysroot Finished release [optimized + debuginfo] target(s) in 0.04s Blocking waiting for file lock on x86_64-bootloader's sysroot Blocking waiting for file lock on x86_64-unknown-linux-gnu's sysroot Blocking waiting for file lock on x86_64-bootloader's sysroot Finished release [optimized + debuginfo] target(s) in 0.05s Blocking waiting for file lock on x86_64-unknown-linux-gnu's sysroot Finished release [optimized + debuginfo] target(s) in 0.04s Finished release [optimized + debuginfo] target(s) in 0.04s ```

Maybe it's possible to first build all tests sequentially and then run them in parallel? I think this could work by iterating twice: In the first iteration we call build::build_impl for each tests so that we get an iterator of test_paths as result. In the second iteration we use par_iter to launch QEMU for each test_path and collect the test results.

[acheronfail]

Maybe it's possible to first build all tests sequentially and then run them in parallel?

That makes things both easier and gives much nicer output. 👌

Output

``` BUILD: test-exception-breakpoint Finished dev [unoptimized + debuginfo] target(s) in 0.05s Building bootloader v0.3.4 Finished release [optimized + debuginfo] target(s) in 0.04s BUILD: test-basic-boot Compiling blog_os v0.2.0 (/Users/cosmotherly/src/blog_os) Finished dev [unoptimized + debuginfo] target(s) in 0.38s Building bootloader v0.3.4 Finished release [optimized + debuginfo] target(s) in 0.04s BUILD: test-panic Finished dev [unoptimized + debuginfo] target(s) in 0.05s Building bootloader v0.3.4 Finished release [optimized + debuginfo] target(s) in 0.04s BUILD: test-exception-double-fault-stack-overflow Finished dev [unoptimized + debuginfo] target(s) in 0.05s Building bootloader v0.3.4 Finished release [optimized + debuginfo] target(s) in 0.04s RUN: test-exception-breakpoint RUN: test-basic-boot RUN: test-panic RUN: test-exception-double-fault-stack-overflow OK: test-panic OK: test-exception-breakpoint OK: test-basic-boot OK: test-exception-double-fault-stack-overflow All tests succeeded. ```

[phil-opp]

Awesome, thanks a lot!

[phil-opp]

Almost forgot to push a new version to crates.io. Version 0.5.7 is out.