In my view the secure boot support provided by mainstream Linux distributions is more about providing installability on systems with secure boot enabled, rather than providing real security benefits.
My reasoning is that while the bootloader and the kernel are signed, the initrd image loaded very early on in boot is not, because it is generated on device. So it provides a convenient way to compromise any system you have physical access to.
The anti evil maid implementation I linked to attempts to mitigate this hole using a TPM. I'm not sure why it isn't on by default but perhaps it's because the implementation has different options that require deciding on a threat model, e.g. setting a TPM password or using an external usb device to store a LUKS key. Here's a good blog post about the anti evil maid implementation that qubes uses(it also works with other distros like Fedora): https://blog.invisiblethings.org/2011/09/07/anti-evil-maid.h...
This blog post contains a good overview of the secure boot status quo along with another potential future fix: https://0pointer.net/blog/brave-new-trusted-boot-world.html