Memory leak persists

[tiagopereira]

There are still memory leaks in RH. They are rather small, but come up as an issue when running a large simulation outside of a supercomputer. This happens also with the lteray binary, so that could narrow down the places where it happens.

[tiagopereira]

Seems that saving tau=1 depths "doubles" the memory leak. Could be important info.

[tiagopereira]

With 4367329 it seems that some memory leak with B projections was closed. Probably not the final say.

[Goobley]

The memory leaks I have found so far are:

• Various temporary allocations used in Profile aren't free'd. My fix is here.
• CollisionRate allocated atom->C but this is normally allocated in readAtom, so just check whether it's non-NULL and free the matrix if there's something there already. See here

I can't directly submit pull requests for these at the moment, due to the other changes I have made to my fork, but if desired I can submit a PR for these. With these changes in place, running rhf1d (including with SOLVE_NE = ITERATION), valgrind reports no leaked memory.

[tiagopereira]

Thanks for looking into this. If you can submit a PR, that would be great and nicer for you to have credit for these changes. I need to double check it with the 1.5D version, but otherwise looks fine!