Add densitymatrix observable at thermal equilibrium

[pablosanjose]

This PR introduces densitymatrix(::GreenSlice,...), which produces a ρ::DensityMatrix over a set of sites. To evaluate the matrix itself at a given temperature and chemical potential, and for a set of system parameters, we do ρ(mu, kBT; params...)

We implement three algorithms in this PR: a generic one that relies on numerical integration on the complex plane of G(ω)*f(ω), another for GreenSolver.Spectrum and another for GreenSolver.KPM at zero temperature. More can be added at a later time.

[codecov-commenter]

Codecov Report

Attention: 12 lines in your changes are missing coverage. Please review.

Comparison is base (27a7e97) 92.66% compared to head (8163698) 92.29%.
Report is 1 commits behind head on master.

Files	Patch %	Lines
src/types.jl	86.66%	4 Missing ⚠️
src/solvers/green/kpm.jl	92.50%	3 Missing ⚠️
src/show.jl	80.00%	2 Missing ⚠️
src/greenfunction.jl	92.85%	1 Missing ⚠️
src/observables.jl	97.67%	1 Missing ⚠️
src/slices.jl	90.00%	1 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #241      +/-   ##
==========================================
- Coverage   92.66%   92.29%   -0.38%     
==========================================
  Files          34       35       +1     
  Lines        5538     5647     +109     
==========================================
+ Hits         5132     5212      +80     
- Misses        406      435      +29     

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

[pablosanjose]

This introduces more refactoring around GreenSlice indexing than I'd like. The reason is that we sometimes need inds and sometimes sites_to_orbs(inds, gs), where inds is some GreenSlice indices. The conversion should only be done once, so we compute it upon construction of a new row/col type GreenIndex living inside GreenSlice. Also, we sometimes want rows or sites_to_orbs(rows, gs) depending on whether rows (and cols) are both contact indices, because then we can get a view of the intracontact GreenSolution that is always computed up-font.

All this functionality should have been split into a separate PR, but then again, this was necessary for densitymatrix with a Spectrum solver, so it landed here. Proper developer discipline is hard! I'll merge this now and remain vigilant for issues from GreenSlice indexing