I checked the latest IEC 61010-1 and IEC 6101-2-030, both are new versions (2017-01-01), replacing the old versions (2010-06-01).
Within these European standards, Safety Classes (e.g. II) and voltage ranges are defined:
The relevant ranges are (>300 <=600V) a.c. rms or dc and (>600 and <= 1000V) a.c. rms or dc.
These ranges require for example different clearances (isolation, jacks, PCB routes,..) and different constant or impulse voltages..
For example, the maximum transient overvoltage in CAT II, for 300..600V is 4kV, for 600..1000V it's 6kV.
The latter value has been increased (from 4kV) in the new revision of IEC 61010-2-030.
Also, the requirements for overvoltage limiting circuits (formerly these were explicitly named: varistors, spark gaps, ceramic capacitors, surge arresters) in the IEC 61010-1 has been increased from only 'suppressing such transients w/o overheating', to additionally 'continue to function properly after the test'.
So, here are my two conclusions:
The standard sets a.c. rms and dc values are defined as being equal in the same voltage ranges, which is not correct.
There is no explicit 440V a.c. range defined anywhere in these standards.
Instead, a.c. peak and d.c should be equivalent regarding transients and overvoltage protection.
Keysight corrected that fault by specifying 600V d.c. and 440V a.c. rms.
Although these standards in the past already defined the same classes and voltage limits, these two above mentioned changes (6kV and further functionality) may not have been given by the momentary design any more.
Most probably by reading the AN 34461A-04, there's a single component affected, maybe the spark tube, or maybe also the PCB design (clearances) does also not withstand these higher requirements.
These transients do not occur during normal use of these instruments in a laboratory / work bench, so that explains the statement, that there is no (or only small) risk in applying 1000Vdc / 750Vac.
I also bet, that older HP DMMs also will not fulfill this standard any more, also other DMM manufacturers will very probably not meet these requirements, if they would read these standards carefully and re-check their instruments.
Frank