06/18/2020 range_tree pretty printer drops high bits of start << shift calculation

[pcd1193182]

** Expected Behavior**
When printing a range tree for a large metaslab, proper offsets should be prrinted for every segment.

** Actual Behavior**
Range segments appeared to increase, then reset back down to the starting value, and this repeated over and over. The last range seg was printed by the pretty printer as [0x324fff4ca00 0x324fff68600) (length 0x1bc00). However, the last range seg (as retrieved by walking the btree directly, and printing out the elements manually) had rs_start = 33552997. When shifted by 9 and added to the ms_start (0x32400000000), it should have been [0x327fff4ca00 0x327fff68600) (length 0x1bc00)

Steps To Reproduce the Problem
Create a pool on disks large enough to contain metaslabs larger than 4 GiB, allocate some space on the pool, and look at the ms_allocatable trees.

Additional Context
It's not clear to me why this occurs; tests in python3's repl don't show similar behavior:

>>> x = 33552997
>>> y = 9
>>> z = 3453153705984
>>> start = (x << y) + z
>>> print("{:x}".format(start))
327fff4ca00

It may have something to do with the way drgn represents integers internally.