With all the discussion that's gone on regarding fitting some model of R-2600 to some modification of the Curtiss model 75, 81, or 87 airframe, I think it's at least fair to argue that investing in developing such an aircraft would have been more worthwhile than the XP-46 project, and a more viable parallel development with the XP-40D and likely initiated significantly earlier than the Merlin P-40 project.
Even down-rating take-off power to better comply with toque and related stability limitations seems like it would've been acceptable if overall performance proved promising once in the air. Though with take-off and low-altitude power being overkill in general, a lighter simplified single-speed R-2600 running only in the high gear ratio may have been worth pursuing. (yes, more complicated logistics initially, and not worth doing unless such an aircraft actually showed promise, but in full-scale production, such an engine should be slightly cheaper to manufacture than the contemporary 2-stage variants on top of being shorter and lighter and at least potentially available sooner than either the 9.6 ratio V-1710 or V-1650-1) All that really depends on how early things get started, though. (it's also somewhat akin to the arguments in favor of the lighter, earlier single-speed BMW 139 over the heavier 801)
I'm also not sure, but with the lower power range in mind, the lighter early models using aluminum crankcases might have had more long-term (or larger scale) production adapted into fighter engines rather than phased out. (and yes I also remember that development of the R-2600 slowed down/became troubled once the US entered the war and its chief designer was barred from the project due to German citizenship, though this is less directly relevant for pre-1940 developments)
Aside from that, the issue of prop clearance and prop RPM is something that was in question (in several other threads as well), and going by the 16:9 reduction gearing of the common 'bomber' R-2600s, you really wouldn't want faster prop speeds than that on a fighter either given the tip-speeds of an 11' prop (akin to that of the P-40D) would already be higher at max continuous 2400 RPM with 16:9 reduction than at 2600 RPM with 2:1 reduction on a V-1710, while take-off RPM ratings would push the V-1710 prop speeds a bit higher (1/2 of 3000 RPM vs 9/16 of 2600 RPM, or 1500 vs 1462.5 RPM at the prop) but still quite close.
Additionally, the props used on the Model 75 appear to have been smaller than the 11' of the P-40D given this document:
http://www.wwiiaircraftperformance.org/P-36/Curtiss_Hawk_75-A_Detail_Specifications.pdf
For the R-1820 Cyclone 9 powered variant:
10 ft 6 in propeller and 1616 rpm rotation rate (16:11 ratio, 2350 rpm R-1820)
10 ft 1 1/2 in prop and 1519 rpm rotation (16:9 reduction, 2700 rpm R-1830)
I did some simple geometry calculations regarding tip speed in flight conditions (taking net velocity of the angular motion from the rotation + perpendicular flight velocity, simple 90 degree pathagorean theorem type stuff), and while those props and rotation rates really wouldn't be cutting it close (except in dives at fairly high altitudes, mostly >400 mph and >20,000 ft) you wouldn't be getting supersonic tip speeds, though cold climates would narrow this margin.
Going up to the 11 ft prop and the higher rotation rates of either the V-1710 or (most) R-2600s could be more problematic though and cutting it somewhat closer, so going with bomber-sized props (or P-47/Corsair sized ones) really wouldn't be an option even aside from problems with torque or (especially) ground clearance.
Also note this issue of supersonic prop speeds would also create some problems with V-1710 overreving, though mostly at high altitude and using 3,400 RPM engine speeds in level flight. (for high alt emergency climb, it might be plausible, but in level flight at 17,000 ft with Mach 1 at around 1050 fps, you'd be hitting supersonic tip speeds at 259 mph) 3200 RPM gives much more practical leeway for boosting top speed, but still runs it fairly close for practical purposes. (just some food for thought in line with Allison potentially performing sufficient testing to allow more regular use of overrev on the E/F series engines, particular with more moderate loads at higher altitudes, somewhat like certain DB 601 models were allowed: though the V-1710 seems to have had a much wider margin for safety in general than the DB engines, or Merlin for that matter; the C-series V-1710 being abused by the AVG and possibly some Colonial units may have also set an additionally conservative president for Allison with their later series engines on top of their limited funding and resource allotment)
Also note that document lists wing guns loading from the top of the wings, with internal ammunition boxes. (it also mentions link and shell retention boxes that may be installed below the ejection chutes of the fuselage guns, so some internet article references to external amunition boxes are likely incorrect, but could be misidentification of detachable link retention boxes for training purposes, retaining spend cartridge cases and belt links: which would better fit the 'cartridge case retainer box' description used on Baugher's website) There's mo mention of external ammo boxes ever being used on the P-36 (or other Hawk models) that I've managed to find on official documents, and this document seems to point to both the single-gun and 2-gun installations in wings used ammunition boxes fitted within the wings and loaded through panels on the upper section of the wing, nothing underneath.
However, in as far as things actually mounted underwing: note alternate armament installation provisions:
1 .50 cal gun below each wing (200 rpg), 1 20 mm oerlikon cannon below each wing (75 rpg), 1 23 mm Madsen cannon below each wing (100 rpg).
The .50 cal option seems most appealing, and might have been a viable trade-off for some speed loss on the P-40B/C/Tomohawk for the added firepower (matching the F2A) prior to the modified 4-gun wings of the Model 87. (and deletion of nose guns entirely) Additionally, the drag added to a radial engine powered fighter would be proportionally smaller than the P-40 itself, and perhaps a more valuable trade-off for relatively limited performance change. (especially for the bulkier R-1820 or a hypothetical R-2600) Incidentally, a 4x .50 cal armed 2-speed Cyclone powered variant might have fared the best in British use as BoB era interceptor, of the Hawk variants available at that time. (fast enough to chase bombers at most altitude in that conflict, much better at damaging fuel tanks and armor, even with ball ammo, thanks to massive hydrostatic shock, and superior roll-rate to anything else in the sky at the time, allowing a totally different sort of out-maneuvering than the simple level turning ability the Hurricane showed; plus all-around lighter controls at high speeds, even if this was curtailed by the higher drag of the engine installation slowing dive acceleration ... perhaps less so when weighed down with armor and tanks wrapped with sealing material; plus good all-around visibility)
Also oddly, there appears to be a substantial weight gain on the R-1830 powered model and I can't work out particular reasons for this.