Fixed bugs in the fixed time output

The changes:

• Dropped the fixed time output member from the classes of the time steppers.
• Removed time stepping update from [TimeStepAlgorithm]::next();
• Made time stepping update take place in the time steppers only after the time step is accepted and the time step sizes from the time steppers of processes are synchronized.

Suggestion: please not use the machine minimum number, std::numeric_limits::min(), as the tolerance for two real number comparison. Such smallest tolerance may give unexpected behavior for some computations under different running environments, e.g a test for the fixed time output works when it runs in a debugger but not in a console.

1. Feature description was added to the changelog, yes.

2. Tests covering your feature were added? Added one more fixed time for output in on existed benchmark. This PR has also been tested with the two benchmarks of 1d-infiltration-RM.prj and 1d-infiltration.prj in another PR for Richards benchmarks (https://github.com/ufz/ogs/pull/2938). @endJunction Please test this PR with that two benchmarks.

3. Any new feature or behavior change was documented? not needed.