It is pretty complex topic, because AR6233 can have several different antenna configuration. It has separated 2.4GHz WiFi RF, Bluetooth RF and 5GHz WiFi RF (actually comes from external Front End Module). How to connect those RF lines to antenna is depends on board design.
While QCA reference design (WB44) uses 2-anntena configuration (One dual-band WiFi antenna + dedicated Bluetooth antenna), the SX-SDCAN used single antenna with diversity. All three RF are combined to single RF line then routed to two dual-band antennas for diversity. Diversity control is optional, so you can also use it as single-antenna module.
The uses of antenna is controlled by the AR6003 (i.e WiFi function). It select 2.4GHz or 5GHz WiFi RF line based on WiFi activity, then let Bluetooth use the antenna for free time. As consequence, WiFi will almost dominant the antenna when you use 2.4GHz WiFI and Bluetooth simultaneously. This is default configuration of WiFi - Bluetooth coexistence behavior. It won't kill each other, but WiFi almost overwhelm the Bluetooth when simultaneously operated.
On SX-SDCAN (and so as SX-SDMAN) the WiFi and Bluetooth functions are connected by 3-wire PTA (Packet Traffic Arbitration) lines for better load sharing between WiFi and Bluetooth. Howver you need special configuration tool from the vendor - "the nasty hack" - to activate the PTA arbitration. I can't tell you much of the detail because I afraid to violate NDA contract with QCA.
The other important fact is that, usage of antenna switch matrix is embedded in the board data file - often refereed as calData file, along with power calibration table. This is another reason you should use the driver (at least the calData file) shipped by the vendor rather than open-source. Standard board data file comes with open-source project will work, but you won't get optimized performance; not only the power calibration would be out of range, but also RF swtich matrix in the calData file may not match with actual module.