[rtl, doc] Consider hardening data_rvalid_i for 2-stage config

[vogelpi]

PR #1968 adds some little extra hardening around data_rvalid_i for the WB-stage configuration used in OpenTitan.

We should probably think whether:

• We can port this extra protection over to the 2-stage configuration.
• How to best document this extra hardening not controllable with the SecureIbex parameter as other countermeasures in case it indeed turns out that the porting cannot be easily done or isn't worth the effort.

For details, refer to #1968 (comment).