Bugfix on basis type instability

[JordiManyer]

The issue here is that TensorValue{2,2,Float64} is actually abstract, the concrete version of it being TensorValue{2,2,Float64,4}. Although the fourth parameter is completely deterministic, Julia does not allow us to define things like struct TensorValue{D1,D2,T} <: MultiValue{D1,D2,T,D1*D2}, which would be ideal. Obviously, the user would always like to use the abstract type, but because the bases construction were just taking whatever type the user provided we ended up with abstract types being used for computations. This is horrible for performance.

To fix this, I have gone through the code and added a way to actually make the user-given types concrete if possible. @Antoinemarteau let me know if you think there is a better way.

Needs to be merged after #1293 .

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.83%. Comparing base (f39eac1) to head (dbef316).
⚠️ Report is 7 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1294      +/-   ##
==========================================
- Coverage   88.83%   88.83%   -0.01%     
==========================================
  Files         227      227              
  Lines       29775    29780       +5     
==========================================
+ Hits        26451    26455       +4     
- Misses       3324     3325       +1     

Flag	Coverage Δ
drivers	39.96% <89.47%> (+<0.01%)	⬆️
extensions	5.12% <22.22%> (-0.02%)	⬇️
unit-adaptivity	40.10% <73.68%> (-0.01%)	⬇️
unit-basics	14.70% <68.42%> (+<0.01%)	⬆️
unit-celldata	21.13% <63.15%> (-0.01%)	⬇️
unit-fespaces-1	32.37% <84.21%> (+<0.01%)	⬆️
unit-fespaces-2	38.84% <73.68%> (+<0.01%)	⬆️
unit-fields	17.62% <100.00%> (+0.04%)	⬆️
unit-geometry	28.77% <57.89%> (-0.01%)	⬇️
unit-multifield	30.88% <63.15%> (-0.01%)	⬇️
unit-odes	28.75% <57.89%> (-0.01%)	⬇️
unit-referencefes	34.25% <100.00%> (+<0.01%)	⬆️
unit-visualization	11.91% <57.89%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.