Domain or watershed data operation about hexwatershed

[changliao1025]

Depression filling must be domain scale.

[changliao1025]

After compset_calculate_flow_accumulation, there are possibly multiple watersheds.
Starting from this point, all the remaining algorithms should consider multiple watersheds.

[changliao1025]

It is also possible that some cells does not belong to any watershed, therefore it is desired to use the domain container so the program won't miss any cell.

[changliao1025]

Once all the cells are processed, we can then group them into watersheds, and output individual watershed as well, just like the pyflowline component.

[changliao1025]

Stream segment topology information maybe added later. And the program needs to sync both domain-based and watershed-based.

[changliao1025]

compset_find_index_by_cellid can only be used at domain.