The graph above is a typical illustration graph of a delayed choice quantumn eraser experiment, and the lower part of this graph includes all parts which are from PS and including PS. There are total 3 typical cases related to delayed choice quantumn eraser experiment. 

First case is exactly as the above graph which is the typical delayed choice quantumn eraser experimemnt, in which the existence of the half mirror lens BSc which erases the path of the marker quantumn, so there will be interference on the D0. 

Second case is: we remove the entire lower part from the prism PS, because we cannot get the path information of the signal and marker photons either, then there will be interference on the D0 as well. 

Third case is: we remove only the BSc from the graph below, we will be able to know the signal photon pass which slit and then our observation or reaching will collapse the signal photons to particles, so there will be NO interference on D0. 

2) The interrupted delayed choice quantumn eraser experiment which is based on the delayed choice quantumn eraser experiment. 

There are two versions of this interrupted delayed choice quantumn eraser experiment: a basic version and an improved version.

For both versions, supposing you have a 10 milliseconds delay after a signal photon reaches D0 and before the marker photon of this signal photon reaches PS/BSc, and supposing you need at least 1000 photons to create an interference, then you can emit 1000 photons one by one in 1ms from the light source to BBO, and then you can do one of two sets steps below to complete this interrupted delayed choice quantumn eraser experiment.

For the basic version, after all the 1000 photons have reached D0, you just need to change the lower part of the above graph either from the first/second case to the third case or from the third case to the first/second case. And thus if this experiment succeeds, the D0 result will not be consistant to the lower part of the below graph.

For the improved version, you need to create an automated process which can determine whether the result of this 1000 photons on the D0 is interference or not in some span like 5ms, and then you can change the lower part of the above graph acording to the result of D0 - if the result of D0 is interference, you change the lower part of the above graph to the third case from the first/second case, and if the result of D0 is not interference, you change the lower part of the above graph to the first/second case from the third case. And thus if this experiment succeeds, the D0 result will not be consistant to the lower part of the above graph.

3) If this interrupted delayed choice quantumn eraser experiment succeed, what could it mean?

If this experiment succeeds which means the result of D0 finally turns out to be not consistant with the lower part of the above graph, I have some guesses below.

First, this universe is prebuilding its possible future, so the time structure of this universe is as graph below.

Second, there are some rules for the possible future of this universe as below. 

In reality and all possible futures of the past & present of this universe, the effects of any interaction should be consistent to the certain fixed laws/rules/principals of this universe.

In this universe, the present reality is capable to interacting with a possible future, and this interacting possible future may not be realized or reached or reality in future, and thus the reality of this universe may lose logistics - losing this interaction's effects in this unrealized possible future out of reality and only have other effects of this interaction in the (present or future) reality of this universe.

In this universe, the past we reached has been certainized which is not able to or not allowed to change, and so any interaction with the certainized past by the present and all possible futures should be consistent to the certain fixed laws/rules/principals of this universe. 

Below is the illustration drawing of the time structure or time field of this universe.