fix syslogd busyloop and enable NULL_TTY

when stdin is /dev/null (e.g. when console=null is on the kernel cmdline) syslogd will call read() over and over and never handle the SIGTERM we send it, pinning a CPU core on every boot.

Provide something on stdin that will always block the read instead, the signal will wake it up and it will properly exit on SIGTERM.

Additionally, follow through on @rdacayan's suggestion and require/enable NULL_TTY on all the kernels where it's available. The driver was only added in 5.2, so having the additional fix in the initramfs is still good for downstream ports.

I ended up actually updating all the kernels here, but we might prefer to leave most/all of the testing kernels alone so we don't flood BPO, I'd appreciate input on that regard.

Fixes: #3269