Enable overridable master algorithm selection for SSP

[markaren]

If accepted, this PR will replace #368, and will close #362 in the process.
It also replaces #370. It's easier to do both at once and will close #369

Instead of completely removing the custom annotation, it can function as a optional fallback similar to how defaultExperiment works in SSP and FMI.

The C++ API stays backwards compatible. The C API now requires 1 additional arguments for the cse_ssp_execution_create function, allowing the startTime defined within the ssp to be optinally used. Otherwise this will function as before.

A new function called cse_ssp_fixed_step_execution_create that takes 4 arguments is also introduced. This function allows the user to assign a custom stepSize value for the algoriithm. In the future this function should be replaced with a generic function taking a cse_algorithm pointer as argument instead. But that`s future work when cse ships with more than one master algorithm.

This PR should make everyone happy, happy, happy. Please have a look at the updated C API doc for the aforementioned function to see if it sounds ok though ..

It should also fix #338, as this PR makes the osp:FixedStepMaster element optional.

[ljamt]

This is a better approach. Let's continue on this PR and close #368 and #370.

Can we increase test coverage of the introduced code?

[markaren]

Can we increase test coverage of the introduced code?

Two new tests added.

[eidekrist]

LGTM