system.tag.queryTagHistory() returning data well outside of bounds

Running 8.1.33
I am retrieving multiple "windows" of data (where the interesting data is), but for some odd reason, each dataset[-1] is HOURS outside of the limits set in the function call. Here is the function call, followed by my diagnostic code.

dsRaw = system.dataset.toPyDataSet(
		system.tag.queryTagHistory(
					paths = tags
					,startDate = start
					,endDate = end
					,columnNames = names
					,noInterpolation = True
					,ignoreBadQuality = True
					)
)
if verbose:
	verboseresult = ["start:{0}".format(start),"end:{0}".format(end),"first:{0}".format(dsRaw[0][0]),"last:{0}".format(dsRaw[-1][0])]
	Statistics.Util.logger("pullCleanHistory() raw results:\n{0}".format("\n".join([3*" " + v for v in verboseresult])))

And here are the results of that diagnostic code:

pullCleanHistory() raw results:
start:Tue Jul 01 08:40:58 EDT 2025
end:Tue Jul 01 08:41:24 EDT 2025
first:Tue Jul 01 08:40:58 EDT 2025
last:Tue Jul 01 13:04:10 EDT 2025

pullCleanHistory() raw results:
start:Tue Jul 01 08:42:24 EDT 2025
end:Tue Jul 01 08:42:49 EDT 2025
first:Tue Jul 01 08:42:24 EDT 2025
last:Tue Jul 01 13:04:10 EDT 2025

pullCleanHistory() raw results:
start:Tue Jul 01 08:43:37 EDT 2025
end:Tue Jul 01 08:44:02 EDT 2025
first:Tue Jul 01 08:43:37 EDT 2025
last:Tue Jul 01 13:04:10 EDT 2025

...etc

While I am showing the first and last data points in the returned dataset, I will note that only the very last data point is outside the limits. I find it odd the same timestamp is returned for all the windows. I hope someone can point me at what I am doing wrong here.
I could perform a check of the last row and just delete it if (when) it is out of bounds, but I'd rather get the return correct than hack the results...if possible. Not like I've never hacked stuff before.

You will need to disable includeBoundingValues if you do not want seed values returned when use interpolation.
You can read up on the options here

Pay attention to

I have noInterpolation set to True and the returnSize should default to -1. I didn't expect any interpolation.

But, I am a good try-er! Unfortunately, setting includeBoundingValues to False did not change the results.

Additionally, I have the tag history set to On-Change and deadband style is Discrete

Strange,
Can you see what tag is coming back with the late date?
Is there an actual stored record for that tag in that partition?

There appear to be valid values for that timestamp, but I was thinking about it and there is also valid data in much earlier timestamps. Just reviewing the results I posted, there are values in the next two minutes, so why it would need to jump hours away is completely beyond me.

Honestly no idea, might want to get support looking over your shoulder on it.
Unless someone from Inductive chimes in with a known bug for that version.