seperate material + stress iteration caches

[kimauth]

Counter suggestion for #42

[codecov-commenter]

Codecov Report

Merging #43 (183eda5) into main (4d791e9) will decrease coverage by 0.75%.
The diff coverage is 86.95%.

@@            Coverage Diff             @@
##             main      #43      +/-   ##
==========================================
- Coverage   92.12%   91.37%   -0.76%     
==========================================
  Files          12       13       +1     
  Lines         432      452      +20     
==========================================
+ Hits          398      413      +15     
- Misses         34       39       +5     

Impacted Files	Coverage Δ
src/wrappers.jl	80.76% <ø> (-1.20%)	⬇️
src/iteration_caches.jl	76.92% <76.92%> (ø)
src/CrystalViscoPlastic/CrystalViscoPlastic.jl	98.57% <100.00%> (+0.02%)	⬆️
src/CrystalViscoPlastic/CrystalViscoPlasticRed.jl	98.64% <100.00%> (+0.01%)	⬆️
src/Plastic.jl	98.36% <100.00%> (+0.02%)	⬆️

---

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 4d791e9...183eda5. Read the comment docs.

[lijas]

I dont really see the difference with this PR and #42. It seems like CacheContainer is just an extra layer, since it is only used when plane stress is needed? So the resulting solver_cache() and plane_stress_cache() just becomes extra boilerplate. Also, the material_response(::Plastic...) seems to also get the PlateStress cache, which it will never need.

Tbh, I think it is okay if we skip making the plane stress iterations allocation free, since it seems to complicate the code to much :P