Key points are highlighted. No comment.
The loco being the heaviest went straight on. Following back from it's final location, allowing for a slight deflection as it slowed to a stop and lost force, and the track it made is clear. The track can be extended backwards to confirm the derailment site. I did this in #18147, #18148, and #18149. There is debris in 18149 marked by yellow arrows and these are zoomed in at #18150.
Following that track of the loco back to where it left the rails:
The yellow zones show the track ballast. Note the precise location in relation to the wall ending, and the white wrecked signal /fencing.
Then look at #18142 and #18143. There is a clear different texture and colour in the track ballast on the near side, the side opposite the derailment. You can see the ballast colour demarcation line clearly in #18152. The ballast likely bounced and loosened here. This is the section of track which vibrated with shock waves from the weight and impact as the heavy loco left the rails and strick the track ties/ballast.
In #18154 the engineer is staring extremely carefully at the track rail/sleeper/ties/ballast in this exact place. Verify location against the perimeter wall ending, and in particular the chain link fence posts.
In #18153 I put 4 x yellow markers from the fence posts across the track numbered 1 to 4. The metal pipe or tree branch at base of the black circle in #18154 is visible to the left of yellow marker line 1, between it and yellow marker line 2. there are track connection welds/bolts in this precise spot on the rails.
That's where it derailed. The reason for the derailment may be the rail, connections, or it may be related to the textural difference in the mound of ballast left of yellow marker line 2 and line 3, where the mound of ballast is slightly lower.
The ballast could have been above rail height, and since swept away by train, other. More likely the track join at this place failed, or the entire rail sank slightly probably by the ballast not being compacted adequately and the track twisting slightly as the near rail depressed under load. This is the beginning of the leftwards curve.
Weight, speed, the curve, and whatever inaccuracy in track geometry existed here temporarily while under load did the rest. If it wasn't the ballast mound directly that did it. I looked carefully at the ballast in these video captures.
There is doubt if the cause of the derailment was just before or just after the rail connections. Intuitively, because the train was moving, I look to the left, or just before the connections for cause. That is at the ballast mound.
The smashed signal/sign/fencing seems to be rather far from the track to have been truck if the train was just derailing there. It could be access ladder/ fittings stripped off a goods wagon or the final portion of perimeter fencing, or lattice/access steps attached to a trackside signal/ speed light. It's location and state is odd in that spot it it was not struck directly by the train because the train should have been only beginning to deviate from the path of the rails at that point.
Bad track laying job ..... most likely.