Remove setitem magic from ring buffer

[cwasicki]

The method is missing wrapping of the index and therefore lacks the essential functionality that the ring buffer should provide. Instead, the update function should be used. So far this method is only used in a very specific unit test.

[llucax]

Seems reasonable to me, my only question is if it doesn't make sense to make __setitem__ use update underneath, to keep the buffer's standard interface, it might come handy at some point if interacting with generic functions, for example from the stdlib.

[cwasicki]

Updated.

my only question is if it doesn't make sense to make setitem use update underneath

In fact both methods are not doing the same. update takes a sample and determines the position in the buffer from the sample's timestamp. It also takes care of gap treatment. With __setitem you specify the position in the underlying buffer directly and bypass that logic, so you could add a value at any position without updating the buffer's internal state to function correctly.

We could still add the setter magic back in future if required, but at the moment I am not sure how exactly that should look like.

[cwasicki]

Resolved conflicts in release notes.

[llucax]

This was removed from the merge queue due to #662. Re-queuing.