Add linting with clang tidy

[alexander-novo]

Was unfortunately unable to find a lint that enforced the MicroGrid namespace, but I'll keep looking.

There are a lot of things that need to be fixed. I'll fix them as a separate pull request.

I turned off the "magic numbers" lint just because I think the mathematical nature of the code will require some magic numbers. I also turned off the "easily swappable parameters" lint because it seemed too burdensome to fix properly, although please check the documentation for the lint and let me know if it's something we should have on.

I don't currently have any "modernize" lints on (lints which encourage using more modern features of C++ over older methods), but I have considered enabling the following:

• modernize-use-emplace
• modernize-use-nullptr
• modernize-use-override
• modernize-use-using

Closes #34

[pelesh]

@alexander-novo, could you explain what this PR accomplishes?

[pelesh]

I ran lint and got ~6,000 lines of suggestions, most of them actually very helpful. Thanks, @alexander-novo!

I can see one possible showstopper: Enforcing variable names to abide by the style guidelines works strict to the letter, but in some cases it would be good to have exceptions. For example, variables Eq and Ed are readily recognized as quadrature and direct potentials, $E_q$ and $E_d$, and make equation code easier to read. If we are to adapt these names to the current rules, we would have to write them as e_q and e_d. Is it possible to (at least) ignore first capital letter in variable names? You would want your variable names in your code to be similar to math symbols in the equations.

@abirchfield, please chime in.

[pelesh]

I turned off the "magic numbers" lint just because I think the mathematical nature of the code will require some magic numbers.

I would say that mathematical nature of the code would be the reason to eliminate all magic numbers. We want all numbers to be defined before we use them. But, I can also see how it can backfire. Lint could catch on ,e.g., $\frac{1}{2}$ in the expression for the kinetic energy $\frac{mv^2}{2}$, so we would need to code it something like

constexpr real_type ONE_HALF = static_cast<real_type>(0.5);

(...)

e_k = ONE_HALF * m * v * v

Perhaps we can live with this?