VRP: upadted problem of trucks starts/ends from different locations

[Westlife1002]

Hi this new question follows the older discussion:
#2620

To make this issue more clear, I am starting another new discussion and attached the full model code which allows helper to look into the root cause.
updated full code: https://gist.github.com/Westlife1002/c6d9c40047723b1581690afcc7ee13b4

In order to enable trucks start from different locations, I have tried to pass two arrays of starts and ends points into the manager. however I have got below issues consistently:

1). problem occured when I change the array of starts from [0,0,0,....] to any others like [1,1,1,....]

Traceback (most recent call last):
  File "C:\Users\user\AppData\Local\Programs\Python\Python36\lib\tkinter\__init__.py", line 1702, in __call__
    return self.func(*args)
  File "C:/Users/user/Dropbox/user/VRP_11.0.py", line 713, in main
    add_time_window_constraints(routing, data, time_callback_index, manager)
  File "C:/Users/user/Dropbox/user/VRP_11.0.py", line 397, in add_time_window_constraints
    time_dimension.CumulVar(index).SetRange(time_window[0], time_window[1])
  File "C:\Users\user\AppData\Roaming\Python\Python36\site-packages\ortools\constraint_solver\pywrapcp.py", line 1098, in SetRange
    return _pywrapcp.IntExpr_SetRange(self, l, u)
Exception: CP Solver fail

I guess this is something wrong related to the time window setup, as a result of changing one start location (deport) to several different locations. If I change the array of starts back to 0, which is the depot index, then the model runs ok. Up to now it's not clear for me to identify the root cause.

2). problem occured when I change the array of ends from [0,0,0, ....] to any others (e.g. [1,1,1,....]
the locations idex and the converted NodeToIndex is different:

location index:  1
node to index:  -1

hence the program stopped with below exit code:
Process finished with exit code -1073741819 (0xC0000005)

Eventually, I want to modify the model to incorportate two scenarios:

1). Allowing arbitrary start and end locations
In my application case, we might purely order trucks from the market. At this point, we do not consider the truck's real-time location. We just need to find the good solution of VRP routes based on the customer locations and demands. Then we base on the VRP solution to find the nearby trucks in the market to pickup the goods. This is similar to the solution here except that we need to add 1-N destination warehouses for unloading: https://developers.google.com/optimization/routing/routing_tasks#allowing-arbitrary-start-and-end-locations
I have problems to fit it into my own model, however.

2). Allowing user-defined multiple start and end locations
such locations might not necessarily be the deport or customer location, but user defined.