Solid construction! Looks like the recorder will fit inside, which is probably one of the more appropriate metrics for determining the correct size.
Getting more technical about it and speculating about the most optimal spacing between microphones based on playback imaging-
The polar pattern of a stereo pair of microphones in combination with the angle between them and spacing between them determines the width of the auditory image heard at playback. This relationship is explained in Michael William's Stereo Zoom paper and in his subsequent work and the work of others. If one knows the polar pattern and angle between a pair of microphones, and there is a specific imaging playback width one wishes to achieve, the appropriate spacing needed to achieve it can be determined.
Refer to the Improved Point At Stacks table linked in my signature. It applies in this case not because this baffled microphone arrangement is intended to be used PAS style for recording an amplified public address system with adjacent pairs of microphones pointed directly at the PA stacks, but because the table solves for the specific Stereo Zoom condition were the resulting image capture angle is the same as the angle between microphones (they typically are not). It would be important in this case if you wished the directional imaging of each adjacent microphone/speaker pair segment to link-up accurately with its adjacent pair along the edges without either too much overlap or a hole in the middle. To be clear- This is not necessary for achieving an immersive sounding recording that will effectively transport the listener back to that acoustic space. It is only important if you want to achieve relatively accurate directional image placement of specific identifiable sounds across in the full 360 degree playback image when played back over 5 speakers arranged in a similar pentagonal arrangement. In other words, don't sweat this unless its important to you to be able to be able to point in the exact direction of a specific croaking frog or squawking bird or whatever.
I'm not sure what polar pattern will be most closely approximated above the frequency where the baffle becomes effective, but lets assume it's essentially cardioid-like in that range. There is a 72-degree angle difference between each adjacent face of the pentagon. Given a pair of cardioids angled 72 degrees apart, the spacing between them which would be needed to achieve an approximate 72 degree wide imaging angle on playback is around 13-15" (30-40cm). At lower frequencies where the baffle becomes ineffective and the polar pattern reverts to omni-directional, the necessary spacing would be more like 30" (77cm) or so between each adjacent pair of microphones. That's a big baffle, and likely too large and unwieldy to be used effectively. Fortunately, given the stated intent of these recordings, it's not critical to get this directional imaging aspect fully optimized.
For reference, with four channels and a square baffle there is a 90 degree angle between microphones, and the dimensions get smaller due to this increased angle. In that case you'd ideally want a spacing of about 8" (20cm) between adjacent cardioids or about 26" (67cm) between adjacent omnis.