Propagate indices in vector indexing of ranges with ranges

[jishnub]

Julia's vector indexing (with OffsetArrays in the mix) follows the rule that axes(r[s]) == axes(s). However, the various getindex methods defined for ranges do not follow this rule, as the result may not be expressed as a range in general if the indices are preserved. OffsetArrays currently carries out extensive type-piracy to "fix" this.

Julia Base:

julia> (1:3)[Base.IdentityUnitRange(2:2)]
2:2

After loading OffsetArrays

julia> (1:3)[Base.IdentityUnitRange(2:2)]
OffsetArrays.IdOffsetRange(values=2:2, indices=2:2)

However this type-piracy of getindex is fragile, since Base (and several packages) dispatch on the first argument of getindex (ie, they define getindex(r::AbstractRange, s)), while OffsetArrays dispatches on the second argument (ie. it defines getindex(r, s::AbstractUnitRange{<:Integer}). This is a minefield of ambiguities, as witnessed recently in JuliaArrays/ArrayInterface.jl#170 (comment).

This PR tries to get around this whole issue by changing how the indices of the indices are propagated. It introduces a function withindices that wraps the result of an indexing operation with the correct axes: withindices(r, axes(s)) has the values of r but the axes of s. Now OffsetArrays only needs to pirate withindices, since the package does not need to compute the values of the result of an indexing operation, only its indices. This implies firstly that fixes to getindex in Base get propagated immediately to OffsetArrays, and secondly OffsetArrays does not introduce ambiguities by dispatching on the second argument.

Currently withindices in Base is a no-op, so it does not change any result and is not a breaking change. The definition of the function really only makes sense if OffsetArrays is loaded, in which case one obtains the correct indices.

Some of the changes in this PR are copied from #41213 as it introduced some fixes for offset ranges. If this PR is merged then the other one might not be necessary. If the other one is merged then this can be rebased on that.

Aside from this there is also a bugfix:

on master:

julia> r = StepRangeLen(Float64(1), Float64(1000), 1000)
1.0:1000.0:999001.0

julia> s = Base.IdentityUnitRange(5:8)
Base.IdentityUnitRange(5:8)

julia> r[s]
ERROR: ArgumentError: StepRangeLen: offset must be in [1,4], got 5

After this PR:

julia> r = StepRangeLen(Float64(1), Float64(1000), 1000)
1.0:1000.0:999001.0

julia> s = Base.IdentityUnitRange(5:8)
Base.IdentityUnitRange(5:8)

julia> r[s]
4001.0:1000.0:7001.0

[jishnub]

Test failure in mpfr seems unrelated. Could someone please review this?

[jishnub]

Test failure in Sockets in tester_linuxaarch64 seems unrelated