use diagonaltensormap

[Jutho]

This starts using the DiagonalTensorMap type in svd and eigenvalue decompositions. It also fixes #181 .

[codecov]

Codecov Report

Attention: Patch coverage is 89.62264% with 11 lines in your changes missing coverage. Please review.

Project coverage is 82.22%. Comparing base (c041bfe) to head (efe4ce0).
Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
src/tensors/factorizations.jl	89.62%	11 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #190      +/-   ##
==========================================
- Coverage   83.16%   82.22%   -0.94%     
==========================================
  Files          42       42              
  Lines        5257     5300      +43     
==========================================
- Hits         4372     4358      -14     
- Misses        885      942      +57     

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

[lkdvos]

I dug a bit into the LinearAlgebra code, for the scalartype they use: https://github.com/JuliaLang/LinearAlgebra.jl/blob/1137b4c7fa8297cef17c4ae0982d7d89d4ab7dd8/src/eigen.jl#L319
I don't know if it makes a difference what the actual implementation is, but I quite like if there is a single function that could be overloaded if needed to control this behavior.

Otherwise looks good to me!

[Jutho]

@lkdvos , is the latest commit more in line but what you had in mind?

[lkdvos]

Yes, exactly! I also like that you factored out the "copy of type" function, looks very clean to me!