Add option for relative tolerance in pytest

[davide-f]

No description provided.

[davide-f]

@AlessandroPampado99 BundleSolver had 1e-4 as default tolerance that is larger than the default 1e-5. Here we propose differentiated tolerances for IB and non-IB and their customization through pytest calls e.g. "pytest --rel=0.0001"

This is ready to merge when CI passes. Subsequently #20

[AlessandroPampado99]

@davide-f tests are not passing for the same reason about tolerance

FAILED test/test_investmentblock.py::test_investment[7: 3n_3c_1gext_1h_1bext_2l.xlsx] - assert 746029920.6 == 736979038.3716856 ± 7.4e+04

comparison failed
Obtained: 746029920.6
Expected: 736979038.3716856 ± 7.4e+04
============ 1 failed, 23 passed, 4 skipped, 24 warnings in 41.48s =============
Error: Process completed with exit code 1.

I'm gonna take a look at it

[davide-f]

@AlessandroPampado99 please share a complete instance that is failing for this case

With the instance, we check if the value is truly random.

We have randomness in the generation of instances right?

In the future we shall improve tests to facilitate to recognize errors in the conversions and errors in smspp.
For example, some tests shall ensure that the smsnetwork has the expected structure and content, regardless of whether it is functional with sms++