implement @time and @timev in terms of @timed

[ericphanson]

closes #47056

I suspect there was some reason this wasn't done in the first place, but I figured opening a PR could be a way to discuss that.

Needs tests still for the new @timed fields.

[IanButterworth]

Sounds good to me. We'd need to check that we're not hitting unexpected compilation due to this change.

[ericphanson]

Do you know a good way to do that?

I tried the example from #48024 (comment):

❯ ./julia --startup-file=no -q
julia> double(x::Real) = 2x
double (generic function with 1 method)

julia> calldouble(container) = double(container[1])
calldouble (generic function with 1 method)

julia> calldouble2(container) = calldouble(container)
calldouble2 (generic function with 1 method)

julia> @time @eval calldouble([1.0])
  0.009445 seconds (2.48 k allocations: 116.594 KiB, 30.16% compilation time) # master
  0.007254 seconds (2.48 k allocations: 116.594 KiB, 30.60% compilation time) # PR
2.0

julia> @time @eval calldouble2(1.0)
  0.002668 seconds (3.81 k allocations: 194.422 KiB, 93.59% compilation time) # master
  0.003196 seconds (3.81 k allocations: 194.422 KiB, 95.07% compilation time) # PR
2.0

which at least looks similar, though I don't know if it means anything.

[IanButterworth]

Needs news for the new fields

[ericphanson]

do we like these field names? we could do compilation_time instead of compile_time since that's how I refer to it in the docstring, though having it shorter seems a bit nicer

[IanButterworth]

compile_time is understandable and shorter is nicer IMO

[IanButterworth]

LGTM. Avoids duplication and the consistency in outputs seems good to me