(Edited according to the discussion with @naturallyInconsistent. The edited part is highlighted in italic.)
We have an experimental bench and we assign a coordinate system $(x,y)$ to it. We shall call $x$ the longitudinal coordinate and $y$ the transverse coordinate. We arrange a laser source at $(-1,0)$, a spectrometer at $(0,0)$, a detector $A$ at $(1,1)$, and another detector $B$ at $(1,-1)$. All of them cannot move longitudinally but can freely move transversely.
First, suppose that the laser source is super strong. Then the two detectors will move away from each other. Each of them will acquire a transverse momentum with the same magnitude but in the opposite direction. The total transverse momentum is always conserved to be $0$.
Now, suppose that the laser source can only emit a single photon at a time.
If we wait for a sufficiently long time to accumulate enough photon emissions and measurement events, we should see the same phenomenon as above.
Then, what happened during a single measurement?
Without loss of generality, let us consider the very first measurement and suppose that detector $A$ detected the photon.
Then there are two possible scenarios:
(I) Detector $A$ acquires a tiny transverse momentum while detector $B$ stays at rest.
This violates momentum conservation.
(II) The two detectors acquire opposite transverse momenta with the same tiny magnitudes.
This violates the locality.
Question: Which scenario describes the actual experimental result after a single measurement? or any other scenario?
Copenhagen's collapse picture seems to support scenario (I), because right before the measurement, according to the evolution equation, the quantum state $\left|{\Psi}\right\rangle$ seems to be a superposition of $\left|{\text{detect, move}}\right\rangle_{A}\otimes\left|{\text{not-detect, rest}}\right\rangle_{B}$ and $\left|{\text{not-detect, rest}}\right\rangle_{A}\otimes\left|{\text{detect, move}}\right\rangle_{B}$. $\left|{\Psi}\right\rangle$ has a zero expectation value of transverse momentum and thus satisfies momentum conservation. The measurement then violates momentum conservation.