[libre-riscv-dev] spike-sv non-default element widths

[Luke Kenneth Casson Leighton]

ok! i think i got it.  took a while.  after redefining RS1, RS2 and
friends to call functions, these obtain the register entry and return,
normally, an XLEN-sized type.

what i realised is: it is perfectly possible now to return a *class* -
which contains the element width.  and, also, operator-overloads are
also possible, which ensures that the absolute perfect answer can be
returned no matter what the combinations of operations.

took long enough to work it out :)

l.