fix test to check process group termination

Summary:
Originally test was implemented only for unix and checked that child process is killed.
In this [PR](facebook#483) test was changed to use a file rather than the PID because Windows reuses PIDs a lot.  But it actually didn't work. The test always passes even if child process is still alive.
This diff returns check for PIDs to ensure that child process is killed. To mitigate PID reuse we now also check for parent PID.

Reviewed By: stepancheg

Differential Revision: D51399236

fbshipit-source-id: 98fe2e25eb639658c04ffd220fa78403cb6e8f4f

Make AnalysisQueryResult struct

Summary: For code navigation. Clear now where `AnalysisQueryResult` is created for example.

Reviewed By: JakobDegen

Differential Revision: D51315032

fbshipit-source-id: 14cf33550ba56d5541a1be154cd2d6dca0106780

fix flakiness in test_artifact_access_time

Summary: Test is reliable on CI but it's flaky on devservers, fix it.

Reviewed By: stepancheg

Differential Revision: D50852868

fbshipit-source-id: 47da11b73212bcf28705dd10f978888ad9686d51

prelude/rust: Process CARGO_MANIFEST_DIR during rustdoc test

Summary:
Rustdoc tests were not getting the `CARGO_MANIFEST_DIR` processing that
we do for all other compilations.

Reviewed By: dtolnay

Differential Revision: D50303301

fbshipit-source-id: 1c3171e7eccc2d1a0461aa93b65ff82ac880c8c0

Augment command execution detail information to test run and test dis…

…covery events

Summary: This diff will attach more information onto test discovery and test run events. Those information will be sent to downstream logging systems.

Reviewed By: get9

Differential Revision: D49745757

fbshipit-source-id: abda9a0ba2823b508d62c731468508b5bde5969a

native python: stop setting PAR_LAUNCH_TIMESTAMP

Summary:
This messes with cli telemetry

Created from CodeHub with https://fburl.com/edit-in-codehub

Reviewed By: itamaro

Differential Revision: D49275018

fbshipit-source-id: c6f2672f3f532fe865a8afb61e6f84af7a798522

Return error from export_as

Summary: `export_as` may construct a type for the typechecker. Type construction may do internal consistency checks. Internal consistency checks may fail. When they fail, better error than panic.

Reviewed By: ianlevesque

Differential Revision: D48857597

fbshipit-source-id: 9339ad0a1e555d4cb5da95852ba6c77d9e4b2f2c

Support coerced attrs value() and type

Summary:
You currently can't do anything with the `coerced_attr` except print them out. Let's implement the basic `type` and `value()` methods to work with coerced attrs, similar to how configured attrs has it.

Coerced attrs aren't really converted to `Value<'v>`s, which is why I think it's ok for this implementation to live in buck2_bxl crate, since that's the only place it would be used.

3 of the coerced attrs types (select, concat, and explicit configured dep) aren't well supported in this implementation, but let's at least support the majority of the coerced attrs for a first pass.

Reviewed By: scottcao

Differential Revision: D48323666

fbshipit-source-id: 16e93ed80b77d0d1804db5919547dd725e52d27a

buck2/oss: use custom %TEMP% in CircleCI on Windows (facebook#371)

Summary:
This helps to eliminate flakiness of doctests on Windows. Sometime they fail with:
```
error: couldn't create a temp dir: The system cannot find the path specified. (os error 3) at path "C:\\Users\\CIRCLE~1.PAC\\AppData\\Local\\Temp\\rustdoctest86bH8a\\rmetawkIVXG"
```

Probably something in CircleCI runner messes with default TEMP dir, causing rustc to fail with this error.

Tested on 15 reruns of CircleCI job, they all succeeded, so I'm pretty confident this helps, although not sure exactly why. Before that, at least few out of 5 were failing.

Pull Request resolved: facebook#371

Reviewed By: ndmitchell

Differential Revision: D47917932

fbshipit-source-id: 5d857e2f947e316e61f65764bb1f70b41bf275b8

rustc_action: always use LF line-endings

Summary: CRLF is a pain. Avoid it.

Reviewed By: shayne-fletcher

Differential Revision: D47553486

fbshipit-source-id: e137526acb72b429f5dd29ff31f30431363783d5