5-term mul! with Diagonal

[jishnub]

This speeds up certain 5-term multiplications involving sparse matrices and a Diagonal, as the operations become O(nnz(A)) from O(N^2).

julia> using SparseArrays, LinearAlgebra, Chairmarks

julia> S = sprand(5_000, 5_000, 0.00001);

julia> S2 = similar(S);

julia> D = Diagonal(axes(S,1));

julia> @b (S2,S,D) mul!(_[1], _[2], _[3], true, false)
190.253 ms # main
7.143 μs # PR

julia> S2 = sprand(5_000, 5_000, 0.00001);

julia> @b (S2,S,D) mul!(_[1], _[2], _[3], 2, 3)
219.410 ms (without a warmup) # main
23.085 μs # PR

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.18%. Comparing base (9548149) to head (a83d27c).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #603      +/-   ##
==========================================
+ Coverage   84.10%   84.18%   +0.07%     
==========================================
  Files          12       12              
  Lines        9192     9249      +57     
==========================================
+ Hits         7731     7786      +55     
- Misses       1461     1463       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.