The new metrics include MAE, MARE, MSE, KGE_D, KGEPRIME_D, CONT_TBL
for deterministic evaluation, and REL_DIAG for probabilistic evaluation.

To be consistent with the other tools in the software stack, the
expected results are now drawn from CSV files that are common across
the stack.

The tests for each metric with output to console are dropped because
too verbose in the scripts. Metrics are only tested against the expected
results in CSV files. The output to console is still tested indirectly
in other unit tests, so this is not a problem.