I do not believe that what @sdyer describes will achieve the correct result. I just tested, and it doesn’t appear to.
The Color Space Transform will be applied in the working space, which if no Input Transform is selected will be linear AWG, converted as if it were ACES2065-1 to ACEScct. So it will have an ACEScct transfer function, and incorrect primaries. Then an AWG to AP1 matrix will be applied to that linear data.
You will get close if you apply two transforms in two node, first an OFX ACES Transform going from ACEScct to ACES, to undo what has happened on the way in, then a Color Space Transform from ARRI Alexa Linear to ACES (AP1) ACEScct. But this will still be missing the chromatic adaptation that is included in a proper IDT.
The best solution would be a DCTL IDT, consisting only of the chromatically adapted AWG to AP0 matrix.
If you are using the Resolve 17 beta, you are able to add custom IDTs using DCTL to the IDT drop-down.