Fix indexing in _mapreducedim for OffsetArrays (#55506)

The destination array was being indexed incorrectly if it had offset
indices. This led to the following on nightly:
```julia
julia> using OffsetArrays

julia> r = 5:100;

julia> a = OffsetVector(r, 2);

julia> sum(a, dims=1)
1-element OffsetArray(::Vector{Int64}, 3:3) with eltype Int64 with indices 3:3:
 0

julia> sum(a)
5040
```
The indexing was marked `@inbounds`, so this was not throwing an error.
This PR also follows #55329 and only marks the indexing operations as
`@inbounds`, omitting the function calls.

---------

Co-authored-by: Matt Bauman <mbauman@juliahub.com>
(cherry picked from commit 3d20a92)

base/reducedim.jl

@@ -261,8 +261,9 @@ function _mapreducedim!(f, op, R::AbstractArray, A::AbstractArrayOrBroadcasted)
         # use mapreduce_impl, which is probably better tuned to achieve higher performance
         nslices = div(length(A), lsiz)
         ibase = first(LinearIndices(A))-1
-        for i = 1:nslices
-            @inbounds R[i] = op(R[i], mapreduce_impl(f, op, A, ibase+1, ibase+lsiz))
+        for i in eachindex(R)
+            r = op(@inbounds(R[i]), mapreduce_impl(f, op, A, ibase+1, ibase+lsiz))
+            @inbounds R[i] = r
             ibase += lsiz
         end
         return R

test/offsetarray.jl

@@ -878,3 +878,10 @@ end
     v = view([1,2,3,4], :)
     @test v[Base.IdentityUnitRange(2:3)] == OffsetArray(2:3, 2:3)
 end
+
+@testset "mapreduce with OffsetRanges" begin
+    r = 5:100
+    a = OffsetArray(r, 2)
+    b = sum(a, dims=1)
+    @test b[begin] == sum(r)
+end