Fix undefined current task variable from #42750

[c42f]

This has caused flaky tests (presumably since #42750 was merged?). For example in

https://buildkite.com/julialang/julia-master/builds/5592#ecf7463c-ed99-41f5-be22-5b665f48c7a1

we have

Error During Test at /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Dates/test/types.jl:277
Test threw exception
Expression: timedwait((()->begin
              false
          end), Second(0); pollint = Millisecond(1)) === :timed_out
UndefVarError: ct not defined
Stacktrace:
 [1] _wait2(c::Base.GenericCondition{Base.Threads.SpinLock}, waiter::Task)
   @ Base ./condition.jl:91
 [2] Timer(cb::Base.var"#timercb#647"{Main.Test27Main_Dates_types.TypesTest.var"#1#2", Channel{Any}, Float64, UInt64}, timeout::Float64; interval::Float64)
   @ Base ./asyncevent.jl:265
 [3] timedwait(testcb::Main.Test27Main_Dates_types.TypesTest.var"#1#2", timeout::Float64; pollint::Float64)
   @ Base ./asyncevent.jl:305
 [4] #timedwait#2
   @ /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Dates/src/types.jl:479 [inlined]
 [5] macro expansion
   @ /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Test/src/Test.jl:479 [inlined]
 [6] macro expansion
   @ /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Dates/test/types.jl:277 [inlined]
 [7] macro expansion
   @ /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Test/src/Test.jl:1347 [inlined]
 [8] top-level scope
   @ /cache/build/amdci5-7/julialang/julia-master/julia-89e19a9c8c/share/julia/stdlib/v1.8/Dates/test/types.jl:277

I'm not sure how better to test this without calling _wait2 directly. Tests for Timer in channels.jl look like they should have covered it (without this they seem to hang on my machine).

[tkf]

How about #41833 (comment) ?

[c42f]

Using current_task() without a temporary makes this code the same as the two other places similar code occurs in task.jl.

(I looked at factoring these occurrences together, but the test isn't exactly the same in the three locations so I figured I'd just fix it in place.)

[tkf]

I see. Makes sense.

[tkf]

Then how about removing the line ct = current_task() in this PR then, so that the patch makes sense in isolation? (i.e., without referring to #41833 for why it's desirable)

[c42f]

Huh? I think this isn't really related to #41833 at all. It's fixing a separate bug introduced in #42750.

The only relation to #41833 is that there were spurious test failures there brought on by this bug.

[c42f]

Oh dear... I see I hit an implicit rebase conflict, and this should go into #41833 after all. Sorry for the noise.