Implement dot_nonrecursive and add fused_map_reduce

[blegat]

JuliaLang/julia#36679 introduced a new function _dot_nonrecursive.
While this function does not exploit mutability, several methods redirect to it so implementing it allows to kill a few birds with one stone.
Our implementation of dot wasn't really correct since we didn't call dot recursively so this PR also fixes that.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@cad04da). Click here to learn what that means.
The diff coverage is 89.18%.

@@            Coverage Diff            @@
##             master      #87   +/-   ##
=========================================
  Coverage          ?   84.82%           
=========================================
  Files             ?       18           
  Lines             ?     1535           
  Branches          ?        0           
=========================================
  Hits              ?     1302           
  Misses            ?      233           
  Partials          ?        0           

Impacted Files	Coverage Δ
src/interface.jl	92.55% <63.63%> (ø)
src/MutableArithmetics.jl	96.66% <100.00%> (ø)
src/dispatch.jl	55.64% <100.00%> (ø)
src/linear_algebra.jl	96.42% <100.00%> (ø)
src/reduce.jl	100.00% <100.00%> (ø)
src/shortcuts.jl	69.23% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cad04da...8d12514. Read the comment docs.

[odow]

Did we try this with the DiffOpt example to see if it works?

[blegat]

No but it's exactly the use case we were discussing in #87 (comment)