tests report:

[npaster]

I have added tests to verify the convergence for the different HHO variants and operators

Where the results are not good:

1. divergence reconstruction: the convergence is suboptimal in 2D with polygonal meshes
2. scalar stabilization: - L2-error does not decrease under 1E-8 (probably a problem of conditioning or round-off error)
   - for the different stabilization, the result is nan with k=3 when the error is small
3. vector stabilization: same remarks than for the scalar stabilization
4. vector symmetric stabilization: same remarks than for the scalar stabilization