Help understanding how the bootloader works

I have the Daisy Seed working perfectly with code running on the internal flash, but obviously 128kB fills up fast.

How does executing from SRAM work? Where is the firmware stored before it is loaded into volatile memory?

Is it possible to execute main code from SRAM and then also load large lookup tables into QSPI_FLASH where they can be accessed with a modicum of speed?

Can/is firmware stored on an SDCARD loaded directly into SRAM for execution, or is it copied to flash somewhere and then loaded into SRAM at runtime?

I assume you have read this documentation page already? → libDaisy/doc/md/_a7_Getting-Started-Daisy-Bootloader.md at master · electro-smith/libDaisy · GitHub

I have and I have perused the linker scripts. I am confused as to where the bootloader is storing the SRAM section in non-volatile memory before it is loaded on boot. Projects compiled with both BOOT_SRAM and BOOT_QSPI both have QSPIFLASH size of 7936K which is 8M - 64K. 64K is not enough memory to store 460K worth of SRAM instructions.