Resolve whether bug leads to a crash or incorrect behavior at runtime

[ChrisTimperley]

Relevant bug: geometry2/1b5fa94/1b5fa94.bug

@gavanderhoorn we're unsure whether this leads to either: (a) a crash at run-time, or (b) using a meaningless/corrupted transform. If the former, then we should go with LOSS-OF-FUNCTIONALITY. If the latter, we should go with BEHAVING-INCORRECTLY.

Originally posted by @ChrisTimperley in #430 (comment)

[ChrisTimperley]

@gavanderhoorn @ipa-hsd I've moved this issue to here.

[gavanderhoorn]

If I understand 1b5fa94 correctly, it would lead to a memory leak.

So (at a high-level): if any Python node which uses TF2 runs for a sufficient amount of time, it will run out of memory and crash.

From the linked PR:

I have found a memory leak in tf2_py.cpp
[..]
Whit this fix the memory leak is fixed.
We have experience this memory leak when having robots working during days.

[gavanderhoorn]

Btw: it looks like ros/geometry2#258 is the issue which reported this (although it was reported by a different user).

Perhaps we should update the .bug?