Add noexcept to all constructors to explicitly state no exception thrown #44

[ThePseudo]

In this PR, I add noexcept to all constructors that are noexcept

[ThePseudo]

@axsaucedo I prepared this PR for #44 but I think there is a design problem to do this. I found this in the failed test because now Memory fails, throws an exception but it gets converted to a segfault in the Tensor class (as it should be).

However, I do not feel right to delete the test for the Tensor/Memory data to be >0 in size or delete the check inside of the tensor.
One easy solution might be not having the Tensor being noexcept, and that would be all.

However, I also do not feel this to be completely right, and as the tensor size should be constant I feel that it should be possible to check it, and (this is really a stretch) I feel that this could be possible at compile time through templates, although I am not sure.

What do you suggest?

[axsaucedo]

Thank you for the contribution @ThePseudo !

Additionally I agree with your comments; to be honest it's been well overdue to properly think about having a better custom exception handling approach, as currently there's quite a few situations where the program exits where a domain-specific exception would be better handled.

One easy solution might be not having the Tensor being noexcept, and that would be all.
For the purpose of this PR I would say we can go forward with this, and then revisit in context of the above; likely this may end up as a bit of a exception to the rule, but let's explore if it can be done (+ wthout a more complex approach).

Based on that let me know and we can merge this one to approach separate.

[ThePseudo]

Hi @axsaucedo, I think it is possible to merge this one. I was experimenting with the other solution and I don't like where it was going: basically it would become a template hell, which would even be ok by me, but everything becoming statically decided does not work for everything (memory usage would be decided at compile time, basically).
My feeling is that I would like to not have exceptions, but in the end it is the most reasonable solution C++ offers us.

[axsaucedo]

My feeling is that I would like to not have exceptions, but in the end it is the most reasonable solution C++ offers us.
Agreed - but for this context specifically, I meant to clean up the areas in the code where we throw a std::runtime_error as to be honest a more specific exception would be more reasonable.

In regards to this PR looks good to go, thanks for the contribution