Fix indexing bug.

[ktbolt]

Current situation

This is the fix for #88.

Code of Conduct & Contributing Guidelines

• I agree to follow the Code of Conduct and Contributing Guidelines.

[MatteoSalvador]

@ktbolt I tried to run the tests locally and all the CFD-related ones fail just after initialization with a double free or corruption (!prev) error. It's kind of interesting that macOS runs smoothly.

[ktbolt]

@MatteoSalvador You run the tests locally on what platform?

I ran pipe3D_RCR on Ubuntu 20 no problem.

[MatteoSalvador]

I ran pipe3D_RCR on Ubuntu 20.04 and I got the error stated before

[ktbolt]

@MatteoSalvador I was not careful and screwed up the indexing of another array. This is why unit tests are so useful.

I found the error by turning on Array index checking. I wonder if we should run the tests with Array indexing checking turned on, even if it slowed down simulations. I could add a flag to the CMake file to enable this.

[MatteoSalvador]

Thanks @ktbolt! I personally agree with turning on array indexing checking in this MR. @mrp089 what do you think?

[mrp089]

I think it makes sense to check indexing in our tests.

This is why unit tests are so useful.

@ktbolt, do you want to add the first one?

[ktbolt]

@mrp089 Implementing unit testing is a lot of work, not sure if we want to expend the resources for this code. Next Stage for sure where we can implement the code for testing (test-driven development).

[MatteoSalvador]

@ktbolt I think that even if all tests are running now, there are still some issues with CFD-related ones.
The solver seems to diverge and the reference solutions are not matched.

[ktbolt]

@MatteoSalvador I saw the divergence, looks like the fix uncovered another bug, looking into that now.

[ktbolt]

@MatteoSalvador The all_fun::integ() functions used to integrate quantities over a face (e.g. all_fun::integ(com_mod, cm_mod, fa, com_mod.Yo, 1, nsd) use a magic number (e.g. 1) as an argument which I misunderstood how it was being used. I modified all those functions to use them the same (awful) way.

[MatteoSalvador]

Thanks @ktbolt! Do you think that this MR will fix the problem @elenasmartinez found with resistance BCs in CFD simulations?

[ktbolt]

@MatteoSalvador This does fix the problem @elenasmartinez found with resistance BCs in CFD.