It seems the 950 hp at 3800 m figure for the Ha 9-II is accurate as it's also posted at several japanese museums with surviving examples of the engine, like:
http://www.hikojour.jp/hikojourB/IMG_65651.jpg
http://www.hikojour.jp/hikojourB/hikojourB0097.html
http://www.hikojour.jp/hikojourB/IMG_656211.jpg
searching "ハ9 エンジン" (Ha9 Engine) will turn up a lot of image search results.
Though it also specifies PS and not HP, so that translates to 937 HP.
The 580 kg figure also appears correct for the later model Ha 9 engines.
Scrolling down to the bottom of the Japanese Wikipedia page on the Ha 9, there's a list of Japanese aircraft engines in a table:
Google Translate
Kawasaki also might not have shortened the stroke of the BMW engine, as BMW may have already done so along with increasing the RPM in their BMW IX development. Very little information seems to be available on the BMW IX itself, so this is purely speculation. (perhaps BMW was less successful with that development than Kawasaki or took greater interest in radial engine development, in either case, the Ha 9's performance and weight seem to put it in the ballpark of the early Jumo 211, which seems like it would've been enough to make the BMW IX in demand had development been similarly successful: and both the Jumo 211A-D and Ha 9 ran unpressurized cooling systems, so radiator drag would've been similar, though the Ha 9 itself was somewhat bulkier in frontal area and never implemented an annular radiator like the 211, which would've complemented that bulk rather well)
On the other hand, given the demand for high performance liquid-cooled engines in the mid/late 1930s, and the number of transports that resorted to using the BMW VI for a time (and some bomber and even fighter prototypes doing so) with both the shortage and limited power of the Jumo 210 and drag and performance limits of the BMW 132 and Bramo 323, and extreme limited availability of the DB 600, it seems like there'd have been a market for the BMW IX if its characteristics were remotely similar to the Ha 9's.
Incidentally, the Ha9's performance range also appears to be quite similar to the Rolls Royce Buzzard, including the maximum power figure of 937 HP (950 PS) being identical to the Buzzard IIIMS. (take-off power was also similarly in the 800-850 HP range)
However, the Ha 2 is listed, but has no article currently, so no further information seems forthcoming there other than its use on the Ki 1 bomber.
The listing in the Japanese Warbirds website:
Google Translate
has common bore and stroke figures to the Hispano 12Y, but it's much heavier than any existing 12Y derivative (555 kg) and doesn't match the performance specs of any existing 12Y engine, plus was used in the Ki 1 in 1932, when the 12Y was first being prototyped. It could be a shortened-stroke derivative of the BMW V
BMW V - Wikipedia or something else entirely and the Warbirds site could have some of the figures in error, including the power.
Google Translate
The Japanese article for the Ki 1 lists the Ha 2 at 700 hp rather than the 940/900 Warbirds has listed, which is very close tot he figure in the English article. The weight listed on warbirds is in the range of the Junkers L55 series V-12, as is the power range, while the dimensions are off. (using an L55 derivative would fit the timeline and Mitsubishi's connection to Junkers with Ki 1 development)
The Ha 4 listed on the Warbirds page doesn't have any mention on Japanese Wikipedia's list of liquid-cooled engines, and also has no mention of the Ha 21 V-12 engine Warbirds has listed.
Also, the Japanese Wiki entry on the Ha 40 (Kawasaki's licensed DB 601) mentions problems with the bearing tolerances not mentioned in the english article along with problems related to the direct-injection fuel system.
Google Translate
bearing
The
tolerance of the surface treatment of the metal bearings used for bearings was about 10 times worse than what is originally required, and the bearings burned well. On the other hand, this machine was polished mechanically, whereas in C 40, due to the shortage of machine tools and its performance problems, compliance with the tolerance of 1 to 2 digits was inevitable due to allowable tolerance compared with the original, It is said that it is because it was the main subject to adjust by
file filing. Also, the fact that the quality of
Kermmet Metal used for bearings was not stable also caused defects
[4] . As a result, breakage of
bearings (bearings) joining the
crankshaft and the main
connecting rod frequently occurred.
Fuel injector
Since the machine tool from Germany became impossible due to the war, the precision of the injection nozzle did not come out, and the passing rate of inspection was only 5%. In addition, problems such as leakage from the flow path due to high pressure applied to the fuel and rupture of the pipe were plagued. Since the fuel injection device itself did not approve the license production by
Bosch who was producing the original
[5] , it was used by remodeling the one produced by
Mitsubishi [6] [* 1] . Mitsubishi has adopted a fuel injection system for Mars of the air-cooled 14 cylinder Martian type two and the like, and there was no problem in reliability in particular. The trouble with Ha 40 and
Atsuta is because both engines are
gasoline direct injection engines , and therefore higher precision is also required for fuel injection equipment, which is said to be surplus in the Japanese technology level at the time .
Lubrication system
A
dry sump system using a
pump and an oil tank was adopted for
lubrication, but if it functions properly, there are many merits but the structure is complicated and the number of manufacturing processes has increased.
Based on that, it seems like at least some problems would've been absent had the Jumo 211F been chosen over the DB 601 for licensed production due to the use of plain bearings compared to the DB 601's roller bearings. (the direct injection system may have still been a difficulty with the 211, though Jumo's design seemed to be more conservative in general and might've been easier to work with)
The DB-600 with its carburetor rather than fuel injection also seems like it would've been a possibility.
The cast block vs discreet cylinder assembly doesn't seem to be a significant factor in any of this in any case which also implies Kawasaki may have been better off simply continuing development of the Ha 9 itself with a follow-on monoblock engine adopting 4 valves per cylinder akin to Soviet development of the AM-34 from the M-17 (itself a licensed BMW VI). It'd likely be somewhat lighter and more compact than the AM-34 given the reduced displacement and existing developments with the Ha 9 as a fighter engine. (there's also some middleground short of the monoblock option, like having reinforcing plates as in the Continental XI-1430 or the cast aluminum cylinder head/camshaft casings like various Fiat V-12 engines used or the similar ones used on the Junkers L88, though some images of L88s appear to be cast-block while others look like individual steel cylinders)
Albeit this picture listed as an L88 appears to be a Mikulin AM38F
Junkers L88 Survivors
vs
File:Mikulin AM-38F.jpg - Wikipedia
So perhaps the L88 was never a monoblock design, as it doesn't appear to be here:
Junkers L88
I'm also not positive the L88 was a 4VPC design, either as Wikipedia lists both it and the L55 as 4VPC, yet it clearly only has single exhaust valves here:
Junkers L55