Adds #33. Note that there is definitely room for future improvement. In particular, we likely want to do something smart when the trailing interval is less than 50% the size of the other intervals in the chunk. Addressing this should be relatively simple when the time comes. There is also room for improvement in how the number of solution intervals per data chunk are computed. Currently I just take the hit and do an early compute - I suspect the correct solution is to move this into the preprocessing. Note that this may be non-trivial.
@bennahugo I am not asking for a review, but just want to make you aware of this.
Adds #33. Note that there is definitely room for future improvement. In particular, we likely want to do something smart when the trailing interval is less than 50% the size of the other intervals in the chunk. Addressing this should be relatively simple when the time comes. There is also room for improvement in how the number of solution intervals per data chunk are computed. Currently I just take the hit and do an early compute - I suspect the correct solution is to move this into the preprocessing. Note that this may be non-trivial.
@bennahugo I am not asking for a review, but just want to make you aware of this.