Property based testing

[SeanBryan51]

The tests for get_fluxsite_comparisons(), and get_fluxsite_tasks() should probably use a property based testing approach rather than testing results against hard coded values.

https://github.com/CABLE-LSM/benchcab/blob/e160bba8324042821d196efb759b534f611327f4/tests/test_fluxsite.py#L394

https://github.com/CABLE-LSM/benchcab/blob/e160bba8324042821d196efb759b534f611327f4/tests/test_fluxsite.py#L422

Properties we could test:

• Test tasks are distinct
• Test number of comparisons is correct ($N = C^{N_r}_2 N_f N_c$ where $C^n_k$ is the binomial coefficient, $N_r$ is the number of realisations, $N_f$ is the number of forcings and $N_c$ is the number of science configurations).
• Test number of tasks is correct (express as a product of the number of branches, forcings, science configurations)

[ccarouge]

Putting it at low priority since there is no development of the fluxsite feature planned in the near future. The tests are working now and will not break unless the fluxsite feature is developed further.