I've been reading the various ST appnotes on this, but I can't find clear answers on whether one can achieve this
1) Disable SWD, so the only way to reload firmware would be with some form of firmware loader (which I have done, via USB MSC FatFS or file upload via HTTP server) and there one can put a secure hash etc on the file being uploaded
2) Prevent bottom 32k of FLASH being read out
3) The secure status being
reversible so if a customer returns a unit it doesn't have to go in the rubbish bin (with various heavy political implications if the customer is a big company with lots of power, and ready to withold payments, etc). ST deny warranty on Level 2 fused devices, which is completely understandable, but they are bigger than anybody else

4) A RAM based boot loader (copied from bottom 32k block to the RAM, for rewriting the whole FLASH) needs to work even when above security features have been set.
AFAICS only
Level 2 security is not reversible by uploading suitably written software.
Item 2) is a bit of an odd one because if you disable SWD, it cannot be read with a debugger anyway, but it needs to be accessible by code executing on the CPU, including the RAM based boot loader. So maybe that bit is pointless.
I've also read that all the time BOOT0/BOOT1 are accessible, somebody can use them to activate ST's boot loader and inject code via e.g. a UART. The method to disable these, by remapping with AF, can apparently be easily bypassed with a carefully timed reset.
I think disabling SWD will be the biggest gain. But it needs to be software-reversible.