Possibly fix #218 - next->_next might be unavailble.

[zeyus]

When cur is deleted, next might become invalid. I've run this test locally and it passes, but I also couldn't get it to fail with the previous implementation, but I see it failing in the test outputs.

If this invokes the test runner we can see if the issue persisits.

[zeyus]

@cboulay for your consideration.

I don't know exactly where it's going wrong, but for example if i close all the inlets in a loop (without deleting them) then the first works, and each subsequent stream_close results in:

(  17.861s) [R_cf_double64   ]      data_receiver.cpp:344    ERR| Stream transmission broke off (Input stream error.); re-connecting...

e.g. for the inlets_10, there'll be that message 9 times.

I don't know if this indicates a problem with the inlets sharing some pointer?

How it is now, the tests pass, and my idea of copying the stream_info is probably not at all related to it.

I'd be happy to make this into a fully fledged PR but I'd prefer to actually understand the problem rather than guess :)

[zeyus]

oh and for the sample destructor, obviously performance is super critical, while .load should give some additional safety (if that is even where the problem really is), I guess there's a slight overhead as well.