Issue #354 analytical damper fix

[sy-cui]

Fix #354.

Instead of the dimensionally inconsistent implementation, three variations of analytical linear dampers are now supported:

1. Uniform damper. The input damping constant uniform_damping_constant $\gamma$ is of dimension 1 / T. Translation and rotation uses the same exponential damping coefficient $\exp(-\gamma t)$. Example:

    simulator.dampen(rod).using(
        AnalyticalLinearDamper,
        uniform_damping_constant=0.1,
        time_step = 1e-4, 
    )

2. Physical damper. The translational and rotational damping constant are in general different, which follows the following formulae

$$m\frac{\partial v}{\partial t} = -\gamma_t v, ~~~~ \frac{J}{e}\frac{\partial \omega}{\partial t} = -\gamma_r \omega$$

such that

$$[\gamma_t] = FL^{-1} T, ~~~~ [\gamma_r] = FLT$$

The users are responsible for choosing the appropriate, dimensionally consistent damping constants for both modes. Example:

    simulator.dampen(rod).using(
        AnalyticalLinearDamper,
        translational_damping_constant=0.1,
        rotational_damping_constant=0.05,
        time_step = 1e-4, 
    )

3. Old implementation (to be deprecated). The syntax is designed to be identical as before, so that client codes do not break. Example:

    simulator.dampen(rod).using(
        AnalyticalLinearDamper,
        damping_constant=0.1,
        time_step = 1e-4, 
    )

Test cases are updated to accommodate the changes. Dissipation module and test cases are moved to dedicated directories.

[skim0119]

Can we keep the changes in dissipation.py?