[assert] Add ASSERT_KNOWN asserts to multiple new outputs

[Razer6]

This adds ASSERT_KNOWN assertions to many outputs of the newly introduced signals.

It also uncovered a bug. The RACL error may be 'x because the internal address hit depends on the incoming TLUL address, which might be 'x at some cases. To fix this, the a_valid signal is factored in to the RACL error computation.

[vogelpi]

Thanks @Razer6 this looks good to me!

Nits:

• Some comments mention that the outputs need to be known after reset. But AFAIK the SVA actually checks they're known all the time.
• One commit message has nbx in the title, I think it should be mbx.

[vogelpi]

CHANGE AUTHORIZED: hw/ip/i2c/rtl/i2c.sv
CHANGE AUTHORIZED: hw/ip/pwm/rtl/pwm.sv
CHANGE AUTHORIZED: hw/ip/spi_host/rtl/spi_host.sv
CHANGE AUTHORIZED: hw/ip/tlul/rtl/tlul_adapter_reg_racl.sv
CHANGE AUTHORIZED: hw/ip/uart/rtl/uart.sv
CHANGE AUTHORIZED: hw/ip/i2c/rtl/i2c_reg_top.sv
CHANGE AUTHORIZED: hw/ip/pwm/rtl/pwm_reg_top.sv
CHANGE AUTHORIZED: hw/ip/uart/rtl/uart_reg_top.sv

This PR only adds missing ASSERT_KNOWN assertions. This doesn't have functional impact.

[Razer6]

• Some comments mention that the outputs need to be known after reset. But AFAIK the SVA actually checks they're known all the time.
• One commit message has nbx in the title, I think it should be mbx.

I changed the mbx commit message. As discussed offline, the comment about after the reset is correct. SVA's are disabled during reset.

[rswarbrick]

CHANGE AUTHORIZED: hw/ip/i2c/rtl/i2c.sv
CHANGE AUTHORIZED: hw/ip/i2c/rtl/i2c_reg_top.sv
CHANGE AUTHORIZED: hw/ip/pwm/rtl/pwm.sv
CHANGE AUTHORIZED: hw/ip/pwm/rtl/pwm_reg_top.sv
CHANGE AUTHORIZED: hw/ip/spi_host/rtl/spi_host.sv
CHANGE AUTHORIZED: hw/ip/tlul/rtl/tlul_adapter_reg_racl.sv
CHANGE AUTHORIZED: hw/ip/uart/rtl/uart.sv
CHANGE AUTHORIZED: hw/ip/uart/rtl/uart_reg_top.sv

This PR is just adding some extra assertions (that we originally missed). No change in functional behaviour.