Avoid use of dimensional tolerances

[andrea-franceschini]

Description As now, LagrangianContactMechanics (#799) uses some dimensional tolerances for displacement jumps and tractions. We should get rid of them, using some internal scaling, based on stiffness, characteristic length or similar ...

[rrsettgast]

@andrea-franceschini @CusiniM Is this done?

[andrea-franceschini]

Based on the same idea behind the stabilization, a fairly robust method for calculating dimensionless tolerances has been proposed, see computeTolerances, however manual corrections are still required for some cases.

If @CusiniM and you agree, I would propose to close this issue now and, if needed, create a new one once we find a case where a revision of the tolerance calculation is required.

[TotoGaz]

Let's follow your advice @andrea-franceschini !