CSub redundant?

[rmn30]

The CSub instruction is currently functionally identical to the RISC-V's sub instruction. The only reason to keep it is to separate pointer operations from integer ones, which could become important in future ISA versions. For example:

• capability addresses might be reduced to less than XLEN width, freeing up capability encoding bits for other uses. In this case CSub may or may not be the same as sub.
• if capability registers were separate from integer ones (split register files).

At the very least we should add some discussion of this to the page in ISA reference.

[nwf-msr]

See also https://github.com/CTSRD-CHERI/cheri-specification/wiki/Tracking-discussion-for-CGetAddr-vs-CToPtr-vs-as-integer-alias-vs-CGetLow