Overhaul of implementation and interface of index manipulations, tensor operations and planar operations

[codecov[bot]]

Codecov Report

Merging #77 (0c4c93e) into master (6c94ed1) will increase coverage by 12.12%. The diff coverage is 79.63%.

@@             Coverage Diff             @@
##           master      #77       +/-   ##
===========================================
+ Coverage   66.31%   78.44%   +12.12%     
===========================================
  Files          40       40               
  Lines        5065     5056        -9     
===========================================
+ Hits         3359     3966      +607     
+ Misses       1706     1090      -616     

Files Changed	Coverage Δ
src/spaces/vectorspaces.jl	91.42% <ø> (ø)
src/tensors/factorizations.jl	95.67% <0.00%> (ø)
src/tensors/linalg.jl	75.72% <0.00%> (ø)
src/tensors/braidingtensor.jl	19.88% <36.36%> (+19.88%)	:arrow_up:
src/spaces/planarspace.jl	40.00% <40.00%> (ø)
src/planar/postprocessors.jl	71.66% <66.66%> (+71.66%)	:arrow_up:
src/tensors/indexmanipulations.jl	72.09% <72.54%> (+6.37%)	:arrow_up:
src/planar/analyzers.jl	72.72% <77.77%> (+72.72%)	:arrow_up:
src/planar/macros.jl	61.03% <90.00%> (+61.03%)	:arrow_up:
src/planar/planaroperations.jl	89.15% <90.90%> (+89.15%)	:arrow_up:
... and 4 more

... and 5 files with indirect coverage changes

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

[Jutho]

Ok, I am merging it like this. The contract! implementation needs a revision (listed as TODO), in order to be able to use different backends. The plan would be:

• If sectortype isa Trivial, simply pas tensors .data to TO.tensorcontract!
• Otherwise, replicate the mul! implementation, but call TO.tensorcontract! on the matrix blocks instead of LinearAlgebra.mul!. TO.tensorcontract! has a dedicated implementation for the case where everything are matrices, but this way one could still have the backend select different matrix multiplication algorithms (e.g. BLAS vs Octavian vs ...)