Changing rs_function and rb_function definitions for user defined ver…

[maritsandstad]

…sions, so obey reasonable constraints and be definable by dictionary with coefficient and timescale vectors

• [x] Tests added
• [x] Documentation added
• [x] Description in CHANGELOG.rst added (single line such as: (`#XX <https://github.com/ciceroOslo/ciceroscm/pull/XX>`_) Added feature which does something)

[benmsanderson]

This works as expected, no issues found.

I do wonder whether we could simplify the dictionaries by having the same length for rs_function coeffs and timscales, and then just have the default parameter for the first mode be large enough to effectively be an infinite timescale.

Added a notebook to test old and new functionality, the following parameter values are pretty close to the old behaviour:

        "rs_function": {
                "coeffs": [0.05, 1.0,0.15,0.15,0.15],
                "timescales": [.8,3,10,80],
            },
        "rb_function": {"coeffs": [0.5, 0.25, 0.25], "timescales": [2.5, 10, 60]},