os/exec: TestFindExecutableVsNoexec failures with `ETXTBSY`

[gopherbot]

#!watchflakes
post <- pkg == "os/exec" && `text file busy`

Issue created automatically to collect these failures.

Example (log):

--- FAIL: TestFindExecutableVsNoexec (0.00s)
    lp_linux_test.go:52: exec: got fork/exec /workdir/tmp/TestFindExecutableVsNoexec867970659/001/program: text file busy, want nil

— watchflakes

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
post <- pkg == "os/exec" && test == "TestFindExecutableVsNoexec"

2022-11-17 03:47 linux-amd64-longtest-race go@fee0ab8b os/exec.TestFindExecutableVsNoexec (log)

--- FAIL: TestFindExecutableVsNoexec (0.00s)
    lp_linux_test.go:52: exec: got fork/exec /workdir/tmp/TestFindExecutableVsNoexec867970659/001/program: text file busy, want nil

— watchflakes

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
post <- pkg == "os/exec" && `text file busy`

2022-12-01 16:39 linux-amd64-nounified go@15e705ea os/exec.TestFindExecutableVsNoexec (log)

--- FAIL: TestFindExecutableVsNoexec (0.02s)
    lp_linux_test.go:52: exec: got fork/exec /workdir/tmp/TestFindExecutableVsNoexec389199221/001/program: text file busy, want nil

— watchflakes

[bcmills]

This is probably a case of #22315.

[gopherbot]

Change https://go.dev/cl/458016 mentions this issue: os/exec: retry ETXTBSY errors in TestFindExecutableVsNoexec

[bcmills]

I made this test parallel in CL 439196, which is what exposed it to #22315 — so this is technically a minor regression.