[enhancement] Refactor of independent_sampling.py

[maxbriel]

This refactors independent_sampling to use the distributions and IMFs for sampling.

This additionally adds:

1. A sampling function per class
2. support for the RNG in inverse_sampler and rejection_sampler
3. Add eccentricity and period classses to distributions.

I have tested this within the context of the unit tests. Not within initial population sampling. However, the norm_pop unit tests contain quite a few tests where the initial samplers are used. These seem to function similarly to before.

This PR does not make any functional changes.

This is the first step to solving issues related the initial sampling, including the difference between setting $m_2$ or $q$ boundaries.

[maxbriel]

Here are distributions. I ran it on the v2.3 branch and this branch.
It's just a bunch of different IMFs, period/separation distributions, and other initial parameters, directly generated using generate_independent_samples.

I didn't see any major differences between the distributions.

population_synthesis_analysis_v2.3.pdf
population_synthesis_analysis_mb_branch.pdf