Worker should kill task handler process and its descendant processes cleanly when timeout

[everpeace]

• fixes Worker doesn't kill taskHandler's subprocesses when timeout #32
• BREAKING CHANGE: This drops windows binary support because this feature requires setpgid
  • I think it doesn't affect much because windows users can WSL2.

$ cat test.sh
#! /bin/bash
sleep 1000

$ ./dist/pftaskqueue start-worker --default-command-timeout 10s --queue-name=test --name=test -- bash ./test.sh
9:18PM INF Start processing a task component=worker processUID=e1543f1f-4bed-463f-a4a7-23c24d2223f0 queueName=test queueUID=9d782669-acb2-4846-89ee-1ae859395c08 taskHandlerCommands=["bash","./test.sh"] taskSpec={"name":"task1","payload":"payload"} taskUID=90eb3dc8-9461-4194-ad9a-ad0f3970ff73 workerName=test workerUID=6e1dca1e-00bc-414f-90ee-71e4bcdf5fbd
9:18PM ERR Task handler process timeout error="context deadline exceeded" component=worker postHooks=[] processUID=e1543f1f-4bed-463f-a4a7-23c24d2223f0 queueName=test queueUID=9d782669-acb2-4846-89ee-1ae859395c08 taskHandlerCommands=["bash","./test.sh"] taskResult={"reason":"Timeout","type":"Failure"} taskSpec={"name":"task1","payload":"payload"} taskUID=90eb3dc8-9461-4194-ad9a-ad0f3970ff73 workerName=test workerUID=6e1dca1e-00bc-414f-90ee-71e4bcdf5fbd
9:18PM INF Completed As Failure because retry exhausted component=redis-backend failureCount=1 operation=RecordFailure queueName=test queueUID=9d782669-acb2-4846-89ee-1ae859395c08 retryLimit=0 workerName=test workerUID=6e1dca1e-00bc-414f-90ee-71e4bcdf5fbd
9:18PM INF Task marked Failure component=worker processUID=e1543f1f-4bed-463f-a4a7-23c24d2223f0 queueName=test queueUID=9d782669-acb2-4846-89ee-1ae859395c08 taskResult={"reason":"Timeout","type":"Failure"} taskSpec={"name":"task1","payload":"payload"} taskUID=90eb3dc8-9461-4194-ad9a-ad0f3970ff73 workerName=test workerUID=6e1dca1e-00bc-414f-90ee-71e4bcdf5fbd
9:18PM INF Queue is empty. retrying in 5 seconds. component=worker detectedQueueEmptyAt=2020-09-29T21:18:47+09:00 exitOnEmpty=false exitOnEmptyGracePeriod=10000 queueName=test queueUID=9d782669-acb2-4846-89ee-1ae859395c08 workerName=test workerUID=6e1dca1e-00bc-414f-90ee-71e4bcdf5fbd
...

# in another shell
# no subprocesses remains!!
$ ps -efj | egrep "[s]leep 1000"